A data analysis method and system for smart education

Abstract

The present invention relates to the field of smart education technology, and in particular to a data analysis method and system for smart education. The method comprises the following steps: obtaining student multimedia usage data and generating independent data sets by grouping by student number; obtaining multimedia usage behavior patterns by feature hierarchical clustering, extracting student test data to generate personal test scores; performing feature matrix interaction analysis and activity integration clustering based on behavior patterns to generate personal learning activity; performing strong and weak learning habit analysis according to learning activity and test score thresholds respectively; performing difference-driven analysis on strong and weak learning habit data to generate beneficial habits and wrong habits data, and realizing intelligent optimization through multimedia management control; the present invention can build a more intelligent, personalized and real-time feedback smart education system through comprehensive and personalized analysis of smart education.

Description

Technical Field

[0001] The present invention relates to the field of smart education technology, and in particular to a data analysis method and system for smart education. Background Art

[0002] In recent years, with the rapid development of information technology, a wave of "smart education" has also been set off in the field of education. In this context, using data analysis methods to explore students' learning behavior patterns, optimize teaching strategies, and improve the quality of education has become an important research direction. Early data analysis methods mainly focused on traditional statistical analysis techniques, such as descriptive statistics, correlation analysis, and regression analysis. These methods can analyze and understand educational data to a certain extent, but the analysis granularity is coarse and cannot deeply explore students' complex learning behavior patterns. The analysis results are also relatively static and superficial, making it difficult to truly understand the deep mechanism behind learning behavior. With the rise of machine learning technology, researchers have begun to try to apply it to educational data analysis to achieve the identification and prediction of learning behavior patterns. These methods have improved the depth and sophistication of analysis to a certain extent, but at the same time have also exposed some new problems, such as weak model interpretability, the need for a large amount of labeled data for training, and poor generalization ability. Summary of the invention

[0003] Based on this, it is necessary to provide a data analysis method and system for smart education to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a data analysis method for smart education includes the following steps:

[0005] Step S1: Acquire student multimedia usage data; group the student multimedia usage data by student ID to generate an independent multimedia usage data set; perform feature hierarchical clustering on the independent multimedia usage data set to obtain multimedia usage behavior patterns; extract student test data from the independent multimedia usage data set to generate individual test scores;

[0006] Step S2: Based on the multimedia usage behavior pattern, feature matrix interaction analysis is performed on the independent multimedia usage data set to obtain an interaction feature matrix; activity integration clustering is performed on the interaction feature matrix to generate personal learning activity;

[0007] Step S3: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and based on the high learning activity and the excellent test score, the independent multimedia usage data set is analyzed for strong subject learning habits to obtain strong subject learning habit data;

[0008] Step S4: when the individual learning activity is less than a preset learning activity threshold and the individual test score is less than a preset test score threshold, the individual learning activity is defined as low learning activity, the individual test score is defined as a lagging test score, and based on the low learning activity and the lagging test score, the weak subject learning habits are derived for the independent multimedia usage data set to obtain weak subject learning habit data;

[0009] Step S5: Perform difference-driven habit analysis on the strong learning habit data and the weak learning habit data to obtain beneficial habit data and incorrect habit data; perform multimedia management and control on the beneficial habit data and the incorrect habit data to achieve intelligent optimization of smart education.

[0010] The present invention constructs a more comprehensive learning data foundation by acquiring information on multiple dimensions of students' multimedia usage data and test scores, performs feature cluster analysis on multimedia usage data, and discovers different multimedia usage behavior patterns of students, laying a foundation for subsequent learning activity evaluation. Based on learning activity and test scores, students are divided into two categories: high activity / excellent and low activity / backward, and their learning habits of strong and weak subjects are analyzed in a targeted manner. By comparing the differences in strong and weak learning habits, beneficial habits and wrong habits are identified, providing a targeted basis for subsequent intelligent optimization. The analysis results of learning habit data are fed back to actual teaching through multimedia management and control, forming a complete closed loop of intelligent education optimization. Data analysis and machine learning technical means are fully utilized, reflecting the technological innovation of intelligent education. The identified beneficial habits and wrong habits are fed back to actual teaching through multimedia management and control, realizing the intelligent optimization of intelligent education. Helping students improve their learning habits and improve their learning effects can comprehensively and accurately understand students' learning behaviors, and through the analysis and optimization of learning habits, the overall effect of intelligent education is improved, the development of intelligent education is promoted, and the intelligent optimization goal of intelligent education is finally achieved. Therefore, the present invention constructs a more intelligent, personalized and real-time feedback smart education system through personalized analysis, comprehensive behavior pattern analysis, dynamic evaluation, strong and weak habit identification, difference-driven habit analysis and multimedia management and control steps, which can better meet the individual needs of students and improve learning effects and education quality.

[0011] In this specification, a data analysis system for smart education is also provided, which is used to execute the data analysis method for smart education as described above. The data analysis system for smart education includes:

[0012] The data acquisition module is used to acquire students' multimedia usage data; group the students' multimedia usage data by student ID to generate independent multimedia usage data sets; perform feature hierarchical clustering on the independent multimedia usage data sets to obtain multimedia usage behavior patterns; extract student test data from the independent multimedia usage data sets to generate individual test scores;

[0013] The activity analysis module is used to perform feature matrix interaction analysis on independent multimedia usage data sets based on multimedia usage behavior patterns to obtain an interaction feature matrix; perform activity integration clustering on the interaction feature matrix to generate personal learning activity;

[0014] A strength identification module is used to define the individual's learning activity as high learning activity and the individual's test score as excellent test score when the individual's learning activity is greater than a preset learning activity threshold and the individual's test score is greater than a preset test score threshold, and to perform a strong subject learning habit analysis on the independent multimedia usage data set based on the high learning activity and the excellent test score to obtain strong subject learning habit data;

[0015] A weakness identification module is used to define the individual's learning activity as low learning activity and the individual's test score as lagging test score when the individual's learning activity is less than a preset learning activity threshold and the individual's test score is less than a preset test score threshold, and to derive weak subject learning habits for the independent multimedia usage data set based on the low learning activity and the lagging test score to obtain weak subject learning habit data;

[0016] The management and control module is used to perform difference-driven habit analysis on strong learning habit data and weak learning habit data to obtain beneficial habit data and incorrect habit data; and to perform multimedia management and control on the beneficial habit data and incorrect habit data to achieve intelligent optimization of smart education.

[0017] The invention is beneficial in that it comprehensively collects students' learning behavior data, ensures the comprehensiveness and accuracy of the data, provides a basis for subsequent analysis, groups data by student number, facilitates personalized analysis, and ensures the pertinence of the data. Cluster analysis is performed on multimedia usage data to identify students' learning behavior patterns and discover potential learning habits and preferences. Students' test scores are extracted and combined with multimedia usage behavior patterns to provide a basis for evaluating learning effects. Through interactive analysis of feature matrices, the associations between different learning behaviors are revealed, providing data support for further optimizing learning strategies. Through cluster analysis, the level of students' learning activity is determined to provide a data basis for personalized teaching. By setting a threshold, students with high learning activity and excellent test scores are accurately identified, which is convenient for analyzing the learning habits of strong subjects for these students. Identifying students' learning habits in strong subjects is convenient for promoting these successful learning strategies and improving overall learning effects. By setting a threshold, students with low learning activity and lagging test scores are accurately identified, which is convenient for deriving learning habits of weak subjects for these students. Identifying students' learning habits in weak subjects is convenient for targeted intervention and counseling, helping students improve their learning methods and improve their grades. By analyzing the differences between strong and weak learning habits, beneficial habit data and wrong habit data are generated to provide a basis for optimizing students' learning behavior. By using beneficial habit data and wrong habit data, the learning behavior of students is monitored and guided in real time through a multimedia management system to achieve intelligent learning optimization and improve the quality of education. Through a detailed analysis of students' learning behavior, individual differences are identified, and personalized learning suggestions and optimization strategies are provided to improve students' learning effects. Using a multimedia management system, students' learning behaviors are monitored in real time, and timely feedback and intervention are provided to ensure that problems in the learning process can be quickly solved. Based on big data analysis, potential learning habits and problems are discovered, and targeted optimization is performed to improve the overall quality of education and student performance. Therefore, the present invention constructs a more intelligent, personalized and real-time feedback smart education system through personalized analysis, comprehensive behavior pattern analysis, dynamic evaluation, strong and weak habit identification, difference-driven habit analysis and multimedia management control steps, which can better meet individual student needs and improve learning effects and education quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of the steps of a data analysis method for smart education;

[0019] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0020] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0021] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0023] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0024] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0025] To achieve this, please refer to Figures 1 to 3 , a data analysis method for smart education, comprising the following steps:

[0026] Step S1: Acquire student multimedia usage data; group the student multimedia usage data by student ID to generate an independent multimedia usage data set; perform feature hierarchical clustering on the independent multimedia usage data set to obtain multimedia usage behavior patterns; extract student test data from the independent multimedia usage data set to generate individual test scores;

[0027] Step S2: Based on the multimedia usage behavior pattern, feature matrix interaction analysis is performed on the independent multimedia usage data set to obtain an interaction feature matrix; activity integration clustering is performed on the interaction feature matrix to generate personal learning activity;

[0028] Step S3: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and based on the high learning activity and the excellent test score, the independent multimedia usage data set is analyzed for strong subject learning habits to obtain strong subject learning habit data;

[0029] Step S4: when the individual learning activity is less than a preset learning activity threshold and the individual test score is less than a preset test score threshold, the individual learning activity is defined as low learning activity, the individual test score is defined as a lagging test score, and based on the low learning activity and the lagging test score, the weak subject learning habits are derived for the independent multimedia usage data set to obtain weak subject learning habit data;

[0030] Step S5: Perform difference-driven habit analysis on the strong learning habit data and the weak learning habit data to obtain beneficial habit data and incorrect habit data; perform multimedia management and control on the beneficial habit data and the incorrect habit data to achieve intelligent optimization of smart education.

[0031] The present invention constructs a more comprehensive learning data foundation by acquiring information on multiple dimensions of students' multimedia usage data and test scores, performs feature cluster analysis on multimedia usage data, and discovers different multimedia usage behavior patterns of students, laying a foundation for subsequent learning activity evaluation. Based on learning activity and test scores, students are divided into two categories: high activity / excellent and low activity / backward, and their learning habits of strong and weak subjects are analyzed in a targeted manner. By comparing the differences in strong and weak learning habits, beneficial habits and wrong habits are identified, providing a targeted basis for subsequent intelligent optimization. The analysis results of learning habit data are fed back to actual teaching through multimedia management and control, forming a complete closed loop of intelligent education optimization. Data analysis and machine learning technical means are fully utilized, reflecting the technological innovation of intelligent education. The identified beneficial habits and wrong habits are fed back to actual teaching through multimedia management and control, realizing the intelligent optimization of intelligent education. Helping students improve their learning habits and improve their learning effects can comprehensively and accurately understand students' learning behaviors, and through the analysis and optimization of learning habits, the overall effect of intelligent education is improved, the development of intelligent education is promoted, and the intelligent optimization goal of intelligent education is finally achieved. Therefore, the present invention constructs a more intelligent, personalized and real-time feedback smart education system through personalized analysis, comprehensive behavior pattern analysis, dynamic evaluation, strong and weak habit identification, difference-driven habit analysis and multimedia management and control steps, which can better meet the individual needs of students and improve learning effects and education quality.

[0032] In the embodiment of the present invention, reference Figure 1The above is a schematic diagram of the steps of a data analysis method for smart education of the present invention. In this example, the data analysis method for smart education includes the following steps:

[0033] Step S1: Acquire student multimedia usage data; group the student multimedia usage data by student ID to generate an independent multimedia usage data set; perform feature hierarchical clustering on the independent multimedia usage data set to obtain multimedia usage behavior patterns; extract student test data from the independent multimedia usage data set to generate individual test scores;

[0034] In an embodiment of the present invention, the interactive data of students when using multimedia teaching resources (such as video, audio, animation) is extracted from the database of the smart education platform, including student ID, access time, access resources, and operation behavior (such as play, pause, fast forward) information, and stored as the original multimedia use data set. This data set records the use of multimedia resources by students in the learning process, provides basic data for subsequent learning behavior analysis, and groups the original multimedia use data set according to the student ID to generate an independent multimedia use data subset for each student. Each subset contains only the multimedia use records of a certain student, which is convenient for subsequent in-depth analysis of the multimedia use behavior of individual students. This operation can utilize the group query function provided by the database management system, or use data analysis tools (such as Excel, Python, R) for data preprocessing and conversion. For each student's independent multimedia use data set, extract characteristic indicators such as access frequency, average access time, access period distribution, and operation behavior characteristics, use these characteristic indicators as input, and use a hierarchical clustering algorithm (such as Ward's method) to classify the student's multimedia use behavior. This process can be implemented using statistical analysis software (such as SPSS, R) or a machine learning library (such as scikit-learn). The clustering results divide students into several types of multimedia usage behavior patterns, providing a basis for subsequent personalized teaching. From the original multimedia usage data set, extract the records of students participating in online tests, including test time, test questions, answer status, and score information, and generate each student's personal test score data. This operation can be achieved through database query or data analysis tools to extract test-related data and calculate individual scores to form a student's personal test score data set. These data provide a basis for subsequent analysis of students' learning effects.

[0035] Step S2: Based on the multimedia usage behavior pattern, feature matrix interaction analysis is performed on the independent multimedia usage data set to obtain an interaction feature matrix; activity integration clustering is performed on the interaction feature matrix to generate personal learning activity;

[0036] In an embodiment of the present invention, the multimedia usage behavior pattern is used to conduct an in-depth analysis of each student's independent multimedia usage data set. First, the characteristic indicators under each usage behavior pattern are extracted, such as access frequency, average access time, access time distribution, and operation behavior, and a characteristic matrix is ​​constructed. Then, the mutual relationship between different students in each characteristic indicator is calculated to form the characteristic matrix interaction analysis results between students. This process can be implemented using data analysis and visualization tools (such as Python's pandas and matplotlib libraries). The obtained interactive feature matrix reflects the similarities and differences in multimedia usage behaviors between students. Based on the aforementioned interactive feature matrix, a clustering algorithm (such as K-means) is used to group students, and students with similar multimedia usage behavior characteristics are classified into the same activity group. On this basis, the indicators of each group of students in terms of access frequency, average access time, and operation behavior are comprehensively considered to calculate the learning activity score of each student. This process can use data mining tools (such as MATLAB, R) or machine learning libraries (such as scikit-learn) to implement cluster analysis and activity evaluation. The personal learning activity information finally obtained provides a basis for teachers to teach students in accordance with their aptitude and provide personalized feedback services to students.

[0037] Step S3: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and based on the high learning activity and the excellent test score, the independent multimedia usage data set is analyzed for strong subject learning habits to obtain strong subject learning habit data;

[0038] In an embodiment of the present invention, a learning activity threshold and a test score threshold are set. The learning activity threshold can be based on the aforementioned clustering results, and a group with higher activity can be selected as a high activity threshold. The test score threshold can be based on the established standards of the school or teacher, and the score line of the excellent grade can be selected as the threshold. When a student's learning activity score is higher than the activity threshold, and his test score is also higher than the score threshold, the student is defined as "high learning activity" and "excellent test score". Based on these two indicators, the student's independent multimedia usage data is deeply analyzed, including access frequency, average access time, access time distribution, and usage function preferences, so as to extract the student's strong learning habits in certain subject areas. This process can be completed using data processing and visualization tools (such as Excel, Tableau). The final strong learning habit data is obtained.

[0039] Step S4: when the individual learning activity is less than a preset learning activity threshold and the individual test score is less than a preset test score threshold, the individual learning activity is defined as low learning activity, the individual test score is defined as a lagging test score, and based on the low learning activity and the lagging test score, the weak subject learning habits are derived for the independent multimedia usage data set to obtain weak subject learning habit data;

[0040] In an embodiment of the present invention, a learning activity threshold and a test score threshold are set. The activity threshold can be selected based on the above-mentioned clustering results, and the group with lower activity can be used as the low activity threshold. The test score threshold can refer to the passing score of the school or teacher as the standard for lagging scores. When a student's learning activity is lower than the set activity threshold, and his test score is also lower than the score threshold, the student's learning activity is defined as "low learning activity" and the test score is defined as "lagging test score". Next, the student's independent multimedia usage data is analyzed, including the characteristics of extremely low access frequency, short average access time, and access periods concentrated in extracurricular time, so as to deduce the student's learning weaknesses in certain subject areas. This process can be completed using data mining and machine learning techniques (such as cluster analysis, association rule mining) with the help of relevant software tools (such as Python, R). The final weak learning habit data is obtained.

[0041] Step S5: Perform difference-driven habit analysis on the strong learning habit data and the weak learning habit data to obtain beneficial habit data and incorrect habit data; perform multimedia management and control on the beneficial habit data and the incorrect habit data to achieve intelligent optimization of smart education.

[0042] In the embodiment of the present invention, the data of strong learning habits and weak learning habits are compared and analyzed. Starting from the dimensions of access frequency, duration, and time period, the beneficial habits and wrong habits of students in the learning process are identified. Beneficial habits include: high learning activity, active and continuous use of multimedia learning resources; good learning effect and excellent test scores. Wrong habits will be manifested as: low learning activity, only concentrated use of learning resources when exams are approaching; poor learning effect, and always lagging behind in test scores. Based on the results of the difference analysis, targeted multimedia management strategies are formulated. For beneficial habits, enhancement measures can be taken, such as optimizing the resource recommendation algorithm to provide students with more accurate and personalized learning resources; strengthening the use of emerging learning tools such as mobile terminals and smart devices to improve the flexibility and convenience of learning. For wrong habits, corrective measures need to be taken. For example, by setting learning tasks and early warning reminders, students can be guided to maintain a good learning state and habits; targeted recommendation of special videos and tutoring article resources can be made to help students fill in knowledge gaps.

[0043] Preferably, step S1 comprises the following steps:

[0044] Step S11: Collecting students' multimedia usage data through the learning management platform;

[0045] Step S12: grouping the students' multimedia usage data by student ID to obtain personal multimedia usage data;

[0046] Step S13: Fill missing values ​​in the personal multimedia usage data to generate an independent multimedia usage data set;

[0047] Step S14: extracting features from the independent multimedia usage data set to generate multimedia feature data;

[0048] Step S15: performing hierarchical clustering on the multimedia feature data to obtain a multimedia usage behavior pattern;

[0049] Step S16: Extract student test data from student multimedia usage data based on multimedia feature data to generate individual test scores.

[0050] The benefit of the present invention is that it can comprehensively and accurately obtain the multimedia usage of students in the learning process, laying a data foundation for subsequent behavioral analysis. Dividing the original data into individual levels is helpful for subsequent targeted analysis of each student's learning behavior. Missing values ​​are filled in personal multimedia usage data to ensure the integrity and reliability of the data and enhance the accuracy of subsequent analysis. Key features that can be used for analysis are extracted from the original data to lay the foundation for learning behavior pattern mining. The typical multimedia usage behavior patterns of students are discovered through clustering algorithms, which helps to deeply understand the learning characteristics of students. Students' test scores can be derived from multimedia usage data, providing an important basis for subsequent learning analysis. Starting from the original data, after cleaning, feature extraction and cluster analysis, students' multimedia usage behavior patterns and test score data are finally obtained, laying a solid data foundation for the construction of a smart education system.

[0051] In an embodiment of the present invention, all operations and activity data of students in a multimedia teaching environment are recorded in real time through a data acquisition module integrated in a learning management platform (LMS). The specific process includes recording the video viewing time, the frequency of use of interactive courseware, the participation in online discussions, and the reading progress of e-books. In the data preprocessing stage, the collected raw multimedia usage data is cleaned and labeled by writing a Python script. The student number is used as a unique identifier to group the data to ensure that each student's multimedia usage record is independent and accurate. The Pandas library is used to group the data and generate a new DataFrame, in which each row of records corresponds to the multimedia usage data of a student. The missing value processing technology in machine learning is used to fill the missing values ​​of personal multimedia usage data. The KNN (K-Nearest Neighbors) algorithm is used to calculate the distance between similar data points and select the nearest neighbor value for filling. First, the data set is converted into a numerical feature matrix, and then the KNNImputer in the Scikit-learn library is applied to fill the missing values ​​to ensure the integrity and continuity of the data set. Feature extraction is performed on the processed independent multimedia usage data set to extract the behavioral characteristics of students in different multimedia usage scenarios. Feature Engineering technology is used to calculate the total usage time, usage frequency of different multimedia resources, and active participation in interaction of each student. Multi-level features are automatically generated using the Featuretools library in Python, and finally a high-dimensional feature matrix is ​​formed, which is called multimedia feature data. Hierarchical clustering algorithm is applied to cluster analysis of multimedia feature data. First, the similarity matrix between feature data is calculated, and then the hierarchy module in the SciPy library is used to perform hierarchical clustering operations, and Ward's method is used to minimize the variance within the cluster. The generated dendrogram is used to determine the optimal number of clusters, and finally the multimedia usage behavior pattern is obtained. Students' multimedia usage data and previous test records are obtained through the learning management platform (LMS). Database query languages ​​(such as SQL) are used to extract students' test records from the LMS database to ensure that the data includes the key fields of student number, test date, and test score. The multimedia feature data is associated with the test record data. The two data sets are merged through the student number field to form a comprehensive data set containing students' multimedia usage behavior and test scores. Use the merge function of the Pandas library to merge the data to ensure that each student's multimedia feature data accurately matches its corresponding test score. Filter the test records and select the three most recent test records as the basis for individual test scores.

[0052] Preferably, step S2 comprises the following steps:

[0053] Step S21: Perform learning behavior screening on the multimedia usage behavior pattern to generate learning behavior data;

[0054] Step S22: performing multimodal fusion on the learning behavior data and the independent multimedia usage data set to obtain a multimodal fusion matrix;

[0055] Step S23: performing feature interaction analysis on the multimodal fusion matrix to obtain an interaction feature matrix;

[0056] Step S24: performing time series analysis on the interaction feature matrix to generate a time feature sequence;

[0057] Step S25: Perform activity integration clustering on the time feature sequence to generate personal learning activity.

[0058] The benefit of the present invention is that behavioral data closely related to learning can be extracted from the original multimedia usage behavior pattern, which is conducive to subsequent learning analysis. Through the fusion of multimodal data, the integration of learning data of different dimensions is achieved, laying the foundation for a more comprehensive learning behavior analysis. Exploring the intrinsic connections and interactive influences between different features will help to understand students' learning mechanisms more deeply. Introducing the analysis of the time dimension can better capture the dynamic changes of students' learning behavior over time. Through activity cluster analysis, the degree of learning participation of different students can be discovered, providing a basis for personalized teaching. The inherent connections and dynamic characteristics of multimodal learning data are fully explored, providing a more comprehensive and accurate student learning portrait for the smart education system, and providing strong support for personalized teaching and learning analysis.

[0059] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0060] Step S21: Perform learning behavior screening on the multimedia usage behavior pattern to generate learning behavior data;

[0061] In an embodiment of the present invention, learning-related behaviors are extracted from multimedia usage behavior patterns through data mining technology. First, the Scikit-learn library in Python is used to preliminarily classify multimedia usage behavior patterns through clustering algorithms such as K-means. Then, based on predefined learning behavior features, such as learning time and the type of educational resources used, the clustering results are filtered. Using a custom filtering function, records that meet the learning behavior features are extracted to form learning behavior data. Finally, the filtered learning behavior data is stored in a relational database and managed and subsequently processed using SQL.

[0062] Step S22: performing multimodal fusion on the learning behavior data and the independent multimedia usage data set to obtain a multimodal fusion matrix;

[0063] In an embodiment of the present invention, multimodal data fusion is performed on the learning behavior data and the independent multimedia usage data set. First, ensure that the two data sets have the same key fields (such as student ID and timestamp) for association. Then, use the merge function in the Pandas library to merge the two data sets by student ID and timestamp. The merged data set contains the students' learning behavior and multimedia usage. Next, use the Numpy library to convert the merged data set into a matrix form to generate a multimodal fusion matrix. The matrix contains the data points of each feature for subsequent feature interaction analysis.

[0064] Step S23: performing feature interaction analysis on the multimodal fusion matrix to obtain an interaction feature matrix;

[0065] In an embodiment of the present invention, an in-depth feature interaction analysis is performed on the multimodal fusion matrix. First, the Pandas and Numpy libraries are used to calculate the correlation coefficient matrix between the features and identify the relationship between the features. Then, the interactive analysis methods in Python, such as factor analysis and principal component analysis (PCA), are used to further explore the hidden feature patterns. The analysis results are integrated to generate an interactive feature matrix. The interactive feature matrix reflects the intrinsic connection and mutual influence between the features, providing data support for subsequent time series analysis.

[0066] Step S24: performing time series analysis on the interaction feature matrix to generate a time feature sequence;

[0067] In an embodiment of the present invention, a time series analysis is performed on the interactive feature matrix to capture the dynamic trend of the feature changes over time. First, the interactive feature matrix is ​​sorted in chronological order using a time series analysis library in Python, such as Statsmodels and Pandas. Then, an ARIMA (autoregressive integrated moving average model) or LSTM (long short-term memory network) time series model is applied to model and predict the sorted data. Through model training and parameter optimization, the significant features in the time series are extracted to generate a time feature series.

[0068] Step S25: Perform activity integration clustering on the time feature sequence to generate personal learning activity.

[0069] In an embodiment of the present invention, activity integrated clustering is performed on the time feature series to evaluate the learning activity of students. First, the time feature series is clustered using an integrated clustering algorithm in the Scikit-learn library, such as DBSCAN (density clustering) or OPTICS (density-based spatial clustering). Appropriate clustering parameters are set by the algorithm to identify different activity patterns. Then, the center and dispersion of each cluster are calculated to determine the learning activity level of each student. Based on the clustering results, personal learning activity data is generated and stored in the database.

[0070] Preferably, step S3 comprises the following steps:

[0071] Step S31: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and the independent multimedia usage data set is matched with the student's strong subjects based on the high learning activity and the excellent test score to obtain the student's learning strengths data;

[0072] Step S32: classify the student's learning strengths data into categories to generate learning strengths category data;

[0073] Step S33: performing strength pattern matching on the multimedia usage behavior pattern based on the learning strength category data to generate a personal strength learning pattern;

[0074] Step S34: recursively analyze the learning rhythm of the individual's strong learning pattern to obtain strong learning habit data.

[0075] The benefit of the present invention is that by setting the learning activity and test score thresholds, the student groups with strong learning ability can be accurately identified, providing a reliable data basis for subsequent personalized teaching. By classifying the students' learning strengths, the learning characteristics and advantages of each student can be described in more detail, providing a reference for the formulation of personalized education strategies. The students' learning strengths categories are matched with their multimedia usage behavior patterns, and the corresponding relationship between learning behavior and subject advantages is established, deepening the understanding of students' learning mechanisms. Through the time series analysis of the individual strengths learning mode, it can be found that the students' learning rhythm and habits in the subjects they are good at are of great guiding significance for teachers to teach students in accordance with their aptitude and adjust teaching methods and progress. Make full use of the students' learning activity, test scores, and learning strengths multi-dimensional data to achieve a comprehensive grasp of students' learning characteristics, learning behavior patterns, and learning habits. These innovations provide more accurate and personalized data analysis support for the smart education system, which helps teachers to formulate teaching strategies that are more in line with students' needs.

[0076] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0077] Step S31: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and the independent multimedia usage data set is matched with the student's strong subjects based on the high learning activity and the excellent test score to obtain the student's learning strengths data;

[0078] In an embodiment of the present invention, the learning activity data of each student is filtered to select records above a preset threshold. Then, the corresponding test scores are screened to select records greater than the preset threshold. Next, SQL query statements are used to cross-match the records that meet the conditions to extract independent multimedia usage data sets of these students. Through data mining technology, the subject usage and learning resource usage frequency in these data sets are analyzed to determine the strong subjects of each student. The analysis results are stored as student learning strengths data for subsequent analysis and processing.

[0079] Step S32: classify the student's learning strengths data into categories to generate learning strengths category data;

[0080] In the embodiment of the present invention, the acquired student learning strengths data is further processed and first classified according to subject categories. The Pandas library in Python is used to group the data, and the data is summarized by subject category through the groupby function. Next, the student strengths data under each subject category is statistically analyzed to calculate the average score and distribution of each category. According to the subject characteristics, the student learning strengths data are divided into different categories, and learning strengths category data are generated.

[0081] Step S33: performing strength pattern matching on the multimedia usage behavior pattern based on the learning strength category data to generate a personal strength learning pattern;

[0082] In the embodiment of the present invention, the learning strengths category data is used to match the multimedia usage behavior pattern. First, a feature matrix containing the learning strengths category data is constructed. Then, a machine learning algorithm (such as KNN or SVM) is used to classify and match the multimedia usage behavior pattern. Through model training, the learning pattern of each student in different subject categories is identified. Then, the matching results are integrated to generate a personal strengths learning pattern.

[0083] Step S34: recursively analyze the learning rhythm of the individual's strong learning pattern to obtain strong learning habit data.

[0084] In an embodiment of the present invention, a data mining algorithm (such as Apriori or FP-Growth) is used to analyze the learning behavior data in the personal strength learning mode to identify the high-frequency action sequences therein. First, the learning behavior data in the personal strength learning mode is converted into an action sequence form to record the specific learning activities of each student in different time periods. Then, a frequent pattern mining algorithm is applied to extract high-frequency action sequence patterns, which represent the behaviors that students often repeat in the learning process. The high-frequency action sequences obtained by mining are stored as learning strength action sequence patterns to facilitate subsequent time period sequence analysis. The learning strength action sequence patterns are grouped and processed by time period. A time series clustering algorithm (such as K-means or DBSCAN) is used to perform cluster analysis on the grouped action sequence data to identify repeated action patterns in different time periods. First, the action sequence patterns are marked by time period to generate a time period sequence data set. Then, a time series clustering algorithm is applied to analyze the data set, cluster similar time period sequences together, and generate time period repeated action data. The data contains the learning actions that students often repeat in a specific time period, providing a data basis for subsequent regularity exploration. Use time series analysis and pattern recognition technology to explore the regularity of repeated action data in time periods. First, arrange the repeated action data in time periods in chronological order and build a time series analysis model. Then, apply pattern recognition algorithms (such as Hidden Markov Model or Dynamic Time Warping) to analyze the data and identify the regular action patterns. Through repeated verification and optimization of these patterns, stable repeated action patterns are extracted. Combine the repeated action patterns with the repeated action data in time periods to conduct in-depth analysis of learning habits. First, compare the repeated action data in time periods with the extracted repeated action patterns to identify the learning behaviors that conform to the rules. Then, use data analysis tools (such as Python's Pandas and NumPy) to perform statistical analysis on the learning behaviors that conform to the rules and calculate the learning frequency and intensity of each student in different time periods. Based on these statistical results, analyze the students' strong learning habits and identify the behavioral patterns that have a positive impact on learning outcomes.

[0085] Preferably, step S34 includes the following steps:

[0086] Step S341: mining the action sequence pattern of the individual's strong learning pattern to obtain the learning strong action sequence pattern;

[0087] Step S342: performing time period sequence clustering processing on the learning strength action sequence pattern to generate time period repeated action data;

[0088] Step S343: exploring the regularity of the repeated action data in the time period to obtain the regularity of the repeated action;

[0089] Step S344: Perform a learning habit analysis on the repeated action data of the time period based on the repeated action rules to generate strong learning habit data.

[0090] The benefit of the present invention is that by mining the action sequence pattern of the individual strength learning pattern, the typical learning behavior pattern of the students in the field of learning strengths can be found, which lays the foundation for the subsequent in-depth analysis of learning habits. Clustering the learning action sequence pattern by time period can identify the action sequence that repeatedly appears in the students' learning strengths, providing a data basis for exploring the law of learning habits. In-depth analysis of the repeated action data of the time period is carried out to dig out the specific repeated action rules of the students in the learning strengths, which provides an important reference for teachers to diagnose and optimize personalized teaching plans. Finally, based on the repeated action rules, a comprehensive analysis of the students' habits in the learning strengths is carried out, and more detailed strength learning habit data is generated, providing a richer personalized analysis basis for the smart education system. The specific behavioral characteristics of students in the learning strengths are deeply excavated and analyzed from multiple levels of learning action sequence, time pattern, and repetition rule, forming a comprehensive learning habit data. These innovations can not only help teachers better understand the learning mechanism of students, but also provide a more accurate personalized service foundation for the smart education system, which has important application value.

[0091] In an embodiment of the present invention, a data mining algorithm (such as Apriori or FP-Growth) is used to analyze the learning behavior data in the personal strength learning mode to identify the high-frequency action sequences therein. First, the learning behavior data in the personal strength learning mode is converted into an action sequence form to record the specific learning activities of each student in different time periods. Then, a frequent pattern mining algorithm is applied to extract high-frequency action sequence patterns, which represent the behaviors that students often repeat in the learning process. The high-frequency action sequences obtained by mining are stored as learning strength action sequence patterns to facilitate subsequent time period sequence analysis. The learning strength action sequence patterns are grouped and processed by time period. A time series clustering algorithm (such as K-means or DBSCAN) is used to perform cluster analysis on the grouped action sequence data to identify repeated action patterns in different time periods. First, the action sequence patterns are marked by time period to generate a time period sequence data set. Then, a time series clustering algorithm is applied to analyze the data set, cluster similar time period sequences together, and generate time period repeated action data. The data contains the learning actions that students often repeat in a specific time period, providing a data basis for subsequent regularity exploration. Use time series analysis and pattern recognition technology to explore the regularity of repeated action data in time periods. First, arrange the repeated action data in time periods in chronological order and build a time series analysis model. Then, apply pattern recognition algorithms (such as Hidden Markov Model or Dynamic Time Warping) to analyze the data and identify the regular action patterns. Through repeated verification and optimization of these patterns, stable repeated action patterns are extracted. Combine the repeated action patterns with the repeated action data in time periods to conduct in-depth analysis of learning habits. First, compare the repeated action data in time periods with the extracted repeated action patterns to identify the learning behaviors that conform to the rules. Then, use data analysis tools (such as Python's Pandas and NumPy) to perform statistical analysis on the learning behaviors that conform to the rules and calculate the learning frequency and intensity of each student in different time periods. Based on these statistical results, analyze the students' strong learning habits and identify the behavioral patterns that have a positive impact on learning outcomes.

[0092] Preferably, step S4 comprises the following steps:

[0093] Step S41: when the individual learning activity is less than a preset learning activity threshold and the individual test score is less than a preset test score threshold, the individual learning activity is defined as low learning activity, the individual test score is defined as a lagging test score, and the independent multimedia usage data set is matched with the student's weak subjects based on the low learning activity and the lagging test score to obtain the student's learning weakness data;

[0094] Step S42: classifying the students' learning weakness data into different categories to generate learning weakness category data;

[0095] Step S43: performing weakness pattern matching on the multimedia usage behavior pattern based on the learning weakness category data to generate a personal weakness learning pattern;

[0096] Step S44: Deducing the learning habits of individual weak learning patterns according to the multimedia usage behavior pattern to obtain weak learning habit data.

[0097] The benefit of the present invention is that by setting learning activity and test score thresholds, the student group with weaker learning performance can be accurately identified, providing a key data basis for targeted teaching and counseling. Categorizing students' learning weaknesses can describe each student's learning difficulties and weak links in more detail, providing a reference for formulating personalized remedial measures. Matching students' learning weakness categories with their multimedia usage behavior patterns helps to discover students' specific learning behavior patterns in weak subjects and provide guidance for teachers to teach students in accordance with their aptitude. Based on the multimedia usage behavior patterns of learning weaknesses, the specific learning habits of students in these weak links are deduced, providing an important basis for the design of targeted counseling programs. The identification, description, correlation analysis and habit deduction of learning weaknesses are comprehensively digitized, providing more accurate analytical support for teachers to diagnose students' learning difficulties and formulate targeted teaching measures. These innovations reflect the comprehensive attention of the smart education system from learning advantages to learning weaknesses, which helps to achieve a more comprehensive and personalized smart education.

[0098] In an embodiment of the present invention, threshold standards for learning activity and test scores are set in a learning management system. Through a data analysis module, students whose learning activity and test scores are both below the threshold are screened out and marked as low learning activity and lagging test scores, respectively. Next, the independent multimedia usage data sets of these students are analyzed, and the weak subjects corresponding to low activity and low test scores are identified using an association rule mining algorithm (such as Apriori). The generated student learning weakness data are imported into a data processing module and classified according to subject categories. Classification algorithms (such as decision trees or random forests) are used to classify weak subjects, and student weakness data are summarized according to different subject categories. Learning weakness category data are generated to record the specific weaknesses of each student in each subject category, providing a data basis for subsequent pattern matching and habit deduction. The learning weakness category data is matched and analyzed with the multimedia usage behavior pattern. First, the multimedia usage behavior data of students in weak subjects is extracted. Then, a pattern matching algorithm (such as K-means or DBSCAN) is applied to analyze these data to identify common behavior patterns existing in weak subjects. These patterns include students' learning behavior habits and specific ways of using multimedia in specific subjects. Generate individual weak learning patterns. Combine multimedia usage behavior patterns with individual weak learning patterns to deduce learning habits. First, use time series analysis methods (such as ARIMA or LSTM) to analyze weak learning patterns and identify regular behaviors in time series. Then, combine the characteristics of multimedia usage behavior patterns to deduce students' learning habits in weak subjects. Through repeated verification and optimization, stable weak learning habit data is obtained to record each student's specific learning habits and behavioral characteristics in weak subjects.

[0099] Preferably, step S44 includes the following steps:

[0100] Step S441: constructing a simulation learning environment for the multimedia usage behavior pattern to obtain a simulated learning environment;

[0101] Step S442: performing weakness learning simulation on the individual's weakness learning mode according to the simulated learning environment to generate a simulated weakness learning process;

[0102] Step S443: monitoring the repeated actions of the simulated weak point learning process to obtain the weak point learning repeated actions;

[0103] Step S444: Analyze the learning habits of individual weak learning patterns based on the weak learning repetitive actions to generate weak learning habit data.

[0104] The benefit of the present invention lies in that by constructing a simulation learning environment based on multimedia usage behavior patterns, a more realistic data basis is provided for subsequent weakness learning simulation, and the reliability of the analysis results is enhanced. By simulating the individual weakness learning pattern in the simulation learning environment, the specific behavior process of students in weakness learning can be restored more accurately, providing a basis for learning habit analysis. Action monitoring is performed for the simulated weakness learning process, and the repeated behavior actions of students when learning weaknesses are identified, which lays the foundation for further exploring the laws of learning habits. Finally, based on the repeated actions in the weakness learning process, the students' weakness learning habits are deeply analyzed, and more detailed weakness learning habit data are generated, providing targeted personalized learning suggestions for the smart education system. Through the construction of the simulation learning environment, the specific behavior process of students in weakness learning is simulated, and the repeated action patterns therein are deeply excavated, and finally the targeted weakness learning habit analysis results are obtained. It can not only help teachers better understand the behavioral characteristics of students in weakness learning, but also provide more accurate personalized tutoring service support for the smart education system.

[0105] In an embodiment of the present invention, a simulation learning environment is constructed based on a multimedia usage behavior pattern. First, students' multimedia usage behavior pattern data is collected, including learning time, usage frequency, and learning content. A virtual learning environment is created using a simulation modeling tool (such as MATLAB Simulink or Unity 3D), which can simulate the multimedia interaction scenarios that students will encounter in actual learning. In the simulation environment, multimedia tools and interactive interfaces are configured to simulate the real learning process and environment. The simulation environment setting should include various learning scenarios and operation steps to ensure that the simulation environment can accurately reproduce the students' multimedia usage behavior. Finally, a simulated learning environment for analysis and testing is obtained. The individual weak learning model is input into the established simulation learning environment. Through the virtual tools in the simulation environment, weak learning simulation is performed according to the learning strategies and methods that students will adopt in actual learning. During the simulation process, the system records the students' learning behavior in weak subjects, including task completion, interaction frequency, and operation steps. The simulated weak learning process generated by the simulation results includes the students' actual operation data in the virtual learning environment, covering learning content, time allocation, and learning strategies. Monitor the learning process of simulated weak points in real time, and use action detection and analysis tools (such as OpenCV in Python or action recognition models in deep learning frameworks) to identify repeated actions in the learning process. The system captures and analyzes all the operation steps of students in the simulated learning process in real time, and determines which operations are repeated through recognition technology. The repeated action data generated by the monitoring system contains the high-frequency operations and repeated steps of students when learning weak points. The data obtained from the repeated action monitoring is input into the learning habit analysis model. Use data mining techniques (such as cluster analysis or association rule analysis) to conduct in-depth analysis of the repeated action data to identify the behavior patterns and habits of students when learning weak points. The analysis model considers the frequency, time distribution and operation content of repeated actions to deduce the students' learning habit data.

[0106] Preferably, step S5 comprises the following steps:

[0107] Step S51: performing a difference comparison between the strong learning habit data and the weak learning habit data to obtain habit difference data;

[0108] Step S52: tracking beneficial habits of the habit difference data according to the strong learning habit data to generate beneficial habit data;

[0109] Step S53: performing wrong habit analysis on the habit difference data according to the weak learning habit data to obtain wrong habit data;

[0110] Step S54: Perform multimedia management and control on beneficial habit data and incorrect habit data to achieve intelligent optimization of smart education.

[0111] The benefit of the present invention lies in that by carefully comparing the strong learning habit data and the weak learning habit data, the specific habit differences of students in the strong and weak learning can be more accurately grasped, providing a basis for subsequent targeted counseling. Based on the strong learning habit data, the habit difference data is tracked for beneficial habits, which helps to find the favorable habits that students should cultivate in the weak learning, and provides guidance for formulating personalized training programs. According to the weak learning habit data, the habit difference data is analyzed for wrong habits, which can identify the unfavorable habit tendencies of students in the weak learning, and provide targeted feedback for teachers to adjust teaching strategies. Finally, through the multimedia management and control of the beneficial habit data and the wrong habit data, the intelligent optimization of students' learning habits is realized, and more optimized personalized learning support is provided for the smart education system. The differences in students' strong and weak learning habits are analyzed in a targeted manner, and the beneficial habits that need to be cultivated and the wrong habits that need to be corrected are excavated, and the intelligent optimization of learning habits is realized through multimedia management. These innovations help teachers to fully understand the learning characteristics of students, and provide a more accurate personalized intervention plan for the smart education system, while improving the teaching effect, it also enhances the students' active learning ability.

[0112] In an embodiment of the present invention, a detailed difference comparison is performed on the strong learning habit data and the weak learning habit data. First, the data analysis tools (such as the Pandas library in Python or the dplyr package in the R language) are used to clean and standardize the strong and weak learning habit data. Then, the two sets of data are compared using statistical analysis methods (such as t-test or variance analysis) to identify the differences. In the process of difference comparison, attention is paid to learning frequency, learning duration, learning strategy and learning resources used. Through comparative analysis, habit difference data is generated to reflect the specific differences between strong and weak learning habits, and the habit difference data is tracked for beneficial habits. First, combined with the strong learning habit data, the habits significantly related to the learning effect are identified through feature selection methods (such as principal component analysis or feature importance ranking). Then, these related habits are classified using clustering algorithms (such as K-means or hierarchical clustering) to determine which habits belong to beneficial habits. During the tracking process, the positive habits in the habit difference data are analyzed in detail to identify their specific contributions to the improvement of academic performance. Finally, beneficial habit data is generated. Conduct wrong habit analysis on habit difference data, based on weak learning habit data. First, analyze the bad learning behaviors in weak learning habit data, and use anomaly detection algorithms (such as isolation forest or LOF algorithm) to identify those habits that have a negative impact on learning effects. Secondly, combine habit difference data to determine the association between these wrong habits and poor learning performance. Compare abnormal behaviors in learning time, learning frequency and learning strategies, and extract common wrong habits. Finally, get wrong habit data. Construct behavior recognition models for beneficial habit data and wrong habit data. First, collect and organize beneficial habit data and wrong habit data to form a training data set. Use data preprocessing technology to clean the data, including removing noise and outliers. Then, select a suitable machine learning algorithm (such as support vector machine (SVM), random forest or neural network) to model the training data. Extract features that are helpful for behavior recognition, such as learning time, operation frequency and learning content type, through feature engineering. Use cross-validation method to evaluate model performance to ensure that the model has high accuracy and stability. Use the constructed learning behavior recognition model to monitor students' multimedia learning behavior in real time. First, the learning behavior recognition model is deployed to a data stream processing platform, such as Apache Kafka or Spark Streaming, to support real-time data processing. The real-time monitoring system accesses the students' multimedia learning behavior data stream, including usage time, interaction frequency, and operation mode. The real-time data is cleaned and formatted by the preprocessing module and then input into the learning behavior recognition model. The model analyzes and classifies the real-time data, identifies the students' current learning behavior patterns, and generates real-time student learning behavior data. When the real-time student learning behavior data is identified as beneficial habit data, beneficial behavior praise measures are taken.First, the system confirms the classification results of the behavior data through the behavior recognition model to determine whether the behavior meets the standards of beneficial habit data. The system automatically generates praise information, such as positive feedback, learning reward points, or personalized learning suggestions. The praise information is pushed to the student account through email, learning management platform notification, or educational application. In order to improve the praise effect, an automated praise mechanism can be set, for example, learning points or badges set in the reward system. When the real-time student learning behavior data is identified as wrong habit data, wrong behavior alarm and behavior misalignment guidance measures are taken. The system first uses the learning behavior recognition model to detect and confirm the wrong habit category of the behavior data. Based on the wrong habit data, detailed alarm information is generated to point out the specific manifestations of the wrong behavior and its negative impact. The alarm information is sent to students through the learning management platform, SMS or application notification, and specific improvement suggestions are provided. In order to correct the wrong behavior, the system uses the beneficial habit data to generate personalized behavior guidance programs, such as recommending effective learning strategies and techniques, and guiding through multimedia teaching resources. Through this comprehensive alarm and guidance mechanism, it aims to correct students' wrong learning behaviors and promote their transformation to beneficial learning habits.

[0113] Preferably, step S54 includes the following steps:

[0114] Step S541: constructing a behavior recognition model for the beneficial habit data and the incorrect habit data to generate a learning behavior recognition model;

[0115] Step S542: using the learning behavior recognition model to perform multimedia learning behavior monitoring and recognition on students to obtain real-time student learning behavior data;

[0116] Step S543: when the real-time student learning behavior data is identified as beneficial habit data, the student is praised for beneficial behavior based on the beneficial habit data to cultivate beneficial learning habits;

[0117] Step S544: When the real-time student learning behavior data is identified as incorrect habit data, the student is alerted of incorrect behavior based on the incorrect habit data, and the real-time student learning behavior data is guided to correct incorrect behavior habits through beneficial habit data.

[0118] The benefit of the present invention lies in that a reliable learning behavior recognition mechanism is established by constructing a behavior recognition model for beneficial habit data and wrong habit data, which provides basic support for subsequent real-time learning behavior monitoring and analysis. Using the constructed learning behavior recognition model, the students' multimedia learning behavior is monitored in real time, and the students' learning behavior data is obtained in real time, providing timely and effective data support for the smart education system. When the student's real-time learning behavior is identified as a beneficial habit, the system immediately gives positive feedback and praise, which is conducive to strengthening the student's good learning behavior and promoting the formation of beneficial learning habits. When the student's real-time learning behavior is identified as an incorrect habit, the system will give an alarm in time and guide and correct it according to the beneficial habit data, so as to effectively correct the student's unfavorable learning behavior and cultivate correct learning habits. A learning behavior recognition model based on beneficial habits and incorrect habits is established, which realizes real-time monitoring and analysis of students' multimedia learning behavior, and takes corresponding feedback and guidance measures for different types of learning behaviors. These innovations help to realize the full tracking and instant optimization of students' learning behavior by the smart education system, greatly improve the pertinence and effect of teaching, and provide strong support for building an intelligent and personalized education system.

[0119] In an embodiment of the present invention, a behavior recognition model is constructed for beneficial habit data and incorrect habit data. First, the beneficial habit data and incorrect habit data are collected and sorted to form a training data set. The data is cleaned using data preprocessing technology, including removing noise and outliers. Next, a suitable machine learning algorithm (such as support vector machine (SVM), random forest or neural network) is selected to model the training data. Features that are helpful for behavior recognition, such as learning time, operation frequency and learning content type, are extracted through feature engineering. The model performance is evaluated by cross-validation method to ensure that the model has high accuracy and stability. The constructed learning behavior recognition model is used to monitor students' multimedia learning behavior in real time. First, the learning behavior recognition model is deployed to a data stream processing platform, such as Apache Kafka or SparkStreaming, to support real-time data processing. The real-time monitoring system accesses the students' multimedia learning behavior data stream, including usage time, interaction frequency and operation mode. The real-time data is cleaned and formatted by the preprocessing module and then input into the learning behavior recognition model. The model analyzes and classifies the real-time data, identifies the students' current learning behavior pattern, and generates real-time student learning behavior data. When real-time student learning behavior data is identified as beneficial habit data, measures to praise beneficial behaviors are taken. First, the system confirms the classification results of the behavior data through the behavior recognition model to determine whether the behavior meets the standards of beneficial habit data. The system automatically generates praise information, such as positive feedback, learning reward points, or personalized learning suggestions. The praise information is pushed to the student account through email, learning management platform notification, or educational application. In order to improve the praise effect, an automated praise mechanism can be set, for example, learning points or badges set in the reward system. When real-time student learning behavior data is identified as wrong habit data, wrong behavior alarm and behavior misalignment guidance measures are taken. The system first uses the learning behavior recognition model to detect and confirm the wrong habit category of the behavior data. Based on the wrong habit data, detailed alarm information is generated to point out the specific manifestations of the wrong behavior and its negative impact. The alarm information is sent to students through the learning management platform, SMS or application notification, and specific improvement suggestions are provided. In order to correct the wrong behavior, the system uses the beneficial habit data to generate personalized behavior guidance plans, such as recommending effective learning strategies and techniques, and guiding through multimedia teaching resources. Through this comprehensive alarm and guidance mechanism, it aims to correct students' wrong learning behaviors and promote their transformation to beneficial learning habits.

[0120] In this specification, a data analysis system for smart education is also provided, which is used to execute the data analysis method for smart education as described above. The data analysis system for smart education includes:

[0121] The data acquisition module is used to acquire students' multimedia usage data; group the students' multimedia usage data by student ID to generate independent multimedia usage data sets; perform feature hierarchical clustering on the independent multimedia usage data sets to obtain multimedia usage behavior patterns; extract student test data from the independent multimedia usage data sets to generate individual test scores;

[0122] The activity analysis module is used to perform feature matrix interaction analysis on independent multimedia usage data sets based on multimedia usage behavior patterns to obtain an interaction feature matrix; perform activity integration clustering on the interaction feature matrix to generate personal learning activity;

[0123] A strength identification module is used to define the individual's learning activity as high learning activity and the individual's test score as excellent test score when the individual's learning activity is greater than a preset learning activity threshold and the individual's test score is greater than a preset test score threshold, and to perform a strong subject learning habit analysis on the independent multimedia usage data set based on the high learning activity and the excellent test score to obtain strong subject learning habit data;

[0124] A weakness identification module is used to define the individual's learning activity as low learning activity and the individual's test score as lagging test score when the individual's learning activity is less than a preset learning activity threshold and the individual's test score is less than a preset test score threshold, and to derive weak subject learning habits for the independent multimedia usage data set based on the low learning activity and the lagging test score to obtain weak subject learning habit data;

[0125] The management and control module is used to perform difference-driven habit analysis on strong learning habit data and weak learning habit data to obtain beneficial habit data and incorrect habit data; and to perform multimedia management and control on the beneficial habit data and incorrect habit data to achieve intelligent optimization of smart education.

[0126] The invention is beneficial in that it comprehensively collects students' learning behavior data, ensures the comprehensiveness and accuracy of the data, provides a basis for subsequent analysis, groups data by student number, facilitates personalized analysis, and ensures the pertinence of the data. Cluster analysis is performed on multimedia usage data to identify students' learning behavior patterns and discover potential learning habits and preferences. Students' test scores are extracted and combined with multimedia usage behavior patterns to provide a basis for evaluating learning effects. Through interactive analysis of feature matrices, the associations between different learning behaviors are revealed, providing data support for further optimizing learning strategies. Through cluster analysis, the level of students' learning activity is determined to provide a data basis for personalized teaching. By setting a threshold, students with high learning activity and excellent test scores are accurately identified, which is convenient for analyzing the learning habits of strong subjects for these students. Identifying students' learning habits in strong subjects is convenient for promoting these successful learning strategies and improving overall learning effects. By setting a threshold, students with low learning activity and lagging test scores are accurately identified, which is convenient for deriving learning habits of weak subjects for these students. Identifying students' learning habits in weak subjects is convenient for targeted intervention and counseling, helping students improve their learning methods and improve their grades. By analyzing the differences between strong and weak learning habits, beneficial habit data and wrong habit data are generated to provide a basis for optimizing students' learning behavior. By using beneficial habit data and wrong habit data, the learning behavior of students is monitored and guided in real time through a multimedia management system to achieve intelligent learning optimization and improve the quality of education. Through a detailed analysis of students' learning behavior, individual differences are identified, and personalized learning suggestions and optimization strategies are provided to improve students' learning effects. Using a multimedia management system, students' learning behaviors are monitored in real time, and timely feedback and intervention are provided to ensure that problems in the learning process can be quickly solved. Based on big data analysis, potential learning habits and problems are discovered, and targeted optimization is performed to improve the overall quality of education and student performance. Therefore, the present invention constructs a more intelligent, personalized and real-time feedback smart education system through personalized analysis, comprehensive behavior pattern analysis, dynamic evaluation, strong and weak habit identification, difference-driven habit analysis and multimedia management control steps, which can better meet individual student needs and improve learning effects and education quality.

[0127] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0128] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A data analysis method for smart education, characterized in that: The following steps are involved: Step S1: Acquire student multimedia usage data; group the student multimedia usage data by student ID to generate an independent multimedia usage data set; Performing feature hierarchical clustering on the independent multimedia usage data set to obtain multimedia usage behavior patterns; extracting student test data from the independent multimedia usage data set to generate individual test scores, wherein step S1 includes the following steps: Step S11: Collecting students' multimedia usage data through the learning management platform; Step S12: grouping the students' multimedia usage data by student ID to obtain personal multimedia usage data; Step S13: Fill missing values ​​in the personal multimedia usage data to generate an independent multimedia usage data set; Step S14: extracting features from the independent multimedia usage data set to generate multimedia feature data; Step S15: performing hierarchical clustering on the multimedia feature data to obtain a multimedia usage behavior pattern; Step S16: extracting student test data from student multimedia usage data based on the multimedia feature data to generate individual test scores; Step S2: Based on the multimedia usage behavior pattern, feature matrix interaction analysis is performed on the independent multimedia usage data set to obtain an interaction feature matrix; activity integration clustering is performed on the interaction feature matrix to generate personal learning activity, wherein step S2 includes the following steps: Step S21: Perform learning behavior screening on the multimedia usage behavior pattern to generate learning behavior data; Step S22: performing multimodal fusion on the learning behavior data and the independent multimedia usage data set to obtain a multimodal fusion matrix; Step S23: performing feature interaction analysis on the multimodal fusion matrix to obtain an interaction feature matrix; Step S24: performing time series analysis on the interaction feature matrix to generate a time feature sequence; Step S25: Perform activity integration clustering on the time feature sequence to generate personal learning activity; Step S3: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and based on the high learning activity and the excellent test score, the independent multimedia usage data set is analyzed for strong subject learning habits to obtain strong subject learning habit data; Step S4: when the individual learning activity is less than a preset learning activity threshold and the individual test score is less than a preset test score threshold, the individual learning activity is defined as low learning activity, the individual test score is defined as a lagging test score, and based on the low learning activity and the lagging test score, the weak subject learning habits are derived for the independent multimedia usage data set to obtain weak subject learning habit data; Step S5: performing difference-driven habit analysis on the strong learning habit data and the weak learning habit data to obtain beneficial habit data and incorrect habit data; performing multimedia management and control on the beneficial habit data and the incorrect habit data to achieve intelligent optimization of smart education, wherein step S5 includes the following steps: Step S51: performing a difference comparison between the strong learning habit data and the weak learning habit data to obtain habit difference data; Step S52: tracking beneficial habits of the habit difference data according to the strong learning habit data to generate beneficial habit data; Step S53: performing wrong habit analysis on the habit difference data according to the weak learning habit data to obtain wrong habit data; Step S54: Perform multimedia management and control on the beneficial habit data and the incorrect habit data to achieve intelligent optimization of smart education, wherein step S54 includes the following steps: Step S541: constructing a behavior recognition model for the beneficial habit data and the incorrect habit data to generate a learning behavior recognition model; Step S542: using the learning behavior recognition model to perform multimedia learning behavior monitoring and recognition on students to obtain real-time student learning behavior data; Step S543: when the real-time student learning behavior data is identified as beneficial habit data, the student is praised for beneficial behavior based on the beneficial habit data to cultivate beneficial learning habits; Step S544: When the real-time student learning behavior data is identified as incorrect habit data, the student is alerted of incorrect behavior based on the incorrect habit data, and the real-time student learning behavior data is guided to correct incorrect behavior habits through beneficial habit data.

2. The data analysis method for smart education according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: When the individual learning activity is greater than a preset learning activity threshold and the individual test score is greater than a preset test score threshold, the individual learning activity is defined as high learning activity, the individual test score is defined as an excellent test score, and the independent multimedia usage data set is matched with the student's strong subjects based on the high learning activity and the excellent test score to obtain the student's learning strengths data; Step S32: classify the student's learning strengths data into categories to generate learning strengths category data; Step S33: performing strength pattern matching on the multimedia usage behavior pattern based on the learning strength category data to generate a personal strength learning pattern; Step S34: recursively analyze the learning rhythm of the individual's strong learning pattern to obtain strong learning habit data.

3. The data analysis method for smart education according to claim 2, characterized in that: Step S34 includes the following steps: Step S341: mining the action sequence pattern of the individual's strong learning pattern to obtain the learning strong action sequence pattern; Step S342: performing time period sequence clustering processing on the learning strength action sequence pattern to generate time period repeated action data; Step S343: exploring the regularity of the repeated action data in the time period to obtain the regularity of the repeated action; Step S344: Perform a learning habit analysis on the repeated action data of the time period based on the repeated action rules to generate strong learning habit data.

4. The data analysis method for smart education according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: when the individual learning activity is less than a preset learning activity threshold and the individual test score is less than a preset test score threshold, the individual learning activity is defined as low learning activity, the individual test score is defined as a lagging test score, and the independent multimedia usage data set is matched with the student's weak subjects based on the low learning activity and the lagging test score to obtain the student's learning weakness data; Step S42: classifying the students' learning weakness data into different categories to generate learning weakness category data; Step S43: performing weakness pattern matching on the multimedia usage behavior pattern based on the learning weakness category data to generate a personal weakness learning pattern; Step S44: Deducing the learning habits of individual weak learning patterns according to the multimedia usage behavior pattern to obtain weak learning habit data.

5. The data analysis method for smart education according to claim 4, characterized in that: Step S44 includes the following steps: Step S441: constructing a simulation learning environment for the multimedia usage behavior pattern to obtain a simulated learning environment; Step S442: performing weakness learning simulation on the individual's weakness learning mode according to the simulated learning environment to generate a simulated weakness learning process; Step S443: monitoring the repeated actions of the simulated weak point learning process to obtain the weak point learning repeated actions; Step S444: Analyze the learning habits of individual weak learning patterns based on the weak learning repetitive actions to generate weak learning habit data.

6. A data analysis system for smart education, characterized in that: Used to execute the data analysis method for smart education according to claim 1, the data analysis system for smart education comprises: The data acquisition module is used to acquire students' multimedia usage data; group the students' multimedia usage data by student ID to generate independent multimedia usage data sets; perform feature hierarchical clustering on the independent multimedia usage data sets to obtain multimedia usage behavior patterns; extract student test data from the independent multimedia usage data sets to generate individual test scores; The activity analysis module is used to perform feature matrix interaction analysis on independent multimedia usage data sets based on multimedia usage behavior patterns to obtain an interaction feature matrix; perform activity integration clustering on the interaction feature matrix to generate personal learning activity; A strength identification module is used to define the individual's learning activity as high learning activity and the individual's test score as excellent test score when the individual's learning activity is greater than a preset learning activity threshold and the individual's test score is greater than a preset test score threshold, and to perform a strong subject learning habit analysis on the independent multimedia usage data set based on the high learning activity and the excellent test score to obtain strong subject learning habit data; A weakness identification module is used to define the individual's learning activity as low learning activity and the individual's test score as lagging test score when the individual's learning activity is less than a preset learning activity threshold and the individual's test score is less than a preset test score threshold, and to derive weak subject learning habits for the independent multimedia usage data set based on the low learning activity and the lagging test score to obtain weak subject learning habit data; The management and control module is used to perform difference-driven habit analysis on strong learning habit data and weak learning habit data to obtain beneficial habit data and incorrect habit data; and to perform multimedia management and control on the beneficial habit data and incorrect habit data to achieve intelligent optimization of smart education.

Images