A comprehensive management method and system for home-school-community collaborative education information data

Abstract

The present invention relates to the field of information management systems and proposes a comprehensive management method and system for information data of collaborative education between home, school and community. The method comprises: collecting data of each dimension of each stage for a number of students; obtaining a baseline value and a correction weight value of attention for each dimension of each student at each stage to obtain attention; obtaining the difference distance between any two students at the same stage and clustering to obtain the cluster to which each student belongs at each stage; obtaining the degree of deviation of each dimension of each student at each stage to obtain the deviation value of each dimension for each student; and performing adaptive PCA dimensionality reduction on each student based on the dimensional vector and deviation value of each dimension of each student to obtain a user profile, thereby completing the comprehensive management of information data of collaborative education between home, school and community. The present invention aims to solve the problem of information loss and erroneous results caused by dimensionality reduction during information management to update user profiles.

Description

Technical Field

[0001] The present invention relates to the field of information management systems, and in particular to a method and system for comprehensive management of home-school-community collaborative education information data. Background Art

[0002] In the 21st century information society, education is no longer confined to the classroom; it has become a comprehensive process encompassing families, schools, and communities. This trend highlights the importance of collaborative education between families, schools, and communities, and necessitates the establishment of effective information and data management systems to accommodate this shift. Today's education generates a vast amount of data, including students' personal information, academic performance, and behavioral records. While this data can be used to improve educational practices, it also presents challenges in data processing and analysis.

[0003] In the current educational environment, families, schools, and communities all play key roles. These entities need to work collaboratively, share information, and regularly update their understanding of students to more effectively support their development. This requires a system that can integrate various data sources, conduct in-depth analysis, and generate reports on student performance, progress, and areas for improvement. Such a system can not only improve the quality and efficiency of education, but also help build and update detailed user profiles for each student. In the process of updating user profiles, it is necessary to perform dimensionality reduction analysis on the collected multidimensional data. However, traditional dimensionality reduction methods, such as the PCA algorithm, can result in information loss in the dimensionality reduction data obtained. Moreover, as time goes by, the multidimensional data that students focus on varies. As a result, the traditional PCA dimensionality reduction method cannot adaptively adjust the eigenvalues ​​according to these differences, which in turn produces erroneous dimensionality reduction analysis results. Summary of the Invention

[0004] The present invention provides a method and system for comprehensive management of educational information data in a collaborative manner between home, school and community, in order to solve the problem of information loss and erroneous results caused by dimensionality reduction during the updating of user portraits in existing information management. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for comprehensive management of home-school-community collaborative education information data, the method comprising the following steps:

[0006] Collect data from several students in each dimension at each stage;

[0007] Based on the distribution of all data of all students in the same stage and the same dimension, the attention benchmark value of each student in each stage and each dimension is obtained; based on the changes in the data of the same dimension in different stages of the same student, the corrected weight value of each dimension in each stage of each student is obtained; based on the corrected weight value and the attention benchmark value, the attention of each student in each dimension in each stage is obtained;

[0008] Based on the differences in data of different students at the same stage and the same dimension, as well as the attention of each student at each stage and each dimension, the difference distance between any two students at the same stage is obtained; clustering is performed based on the difference distance to obtain the cluster to which each student belongs at each stage;

[0009] Based on the adaptive threshold of each dimension of each cluster and the data of each dimension of each student at each stage, the deviation degree of each dimension of each student at each stage is obtained; based on the deviation degree of the same dimension of each student at different stages, the deviation value of each dimension of each student is obtained;

[0010] According to the dimension vector and deviation value of each dimension of each student, adaptive PCA dimensionality reduction is performed on each student to obtain a user portrait, which is stored in the information data comprehensive management platform for collaborative education between home, school and community.

[0011] Furthermore, the attention benchmark value of each dimension of each student at each stage is obtained in the following specific method:

[0012] For the lth dimension of the i-th stage, obtain the data of all students in this dimension in this stage as the data set of this dimension in this stage, obtain the frequency of occurrence of each data in the data set, calculate the variance based on the frequency of occurrence, and record the result as the frequency variance of this dimension in this stage;

[0013] Calculate the absolute value of the difference between any two data in the data set, take the absolute value of the difference as the distance between the two data, start extracting from the minimum value of all distances, and obtain several data pairs without duplication; the baseline value α of the attention of the jth student in the lth dimension in the i-th stage i,l The calculation method of (j) is:

[0014]

[0015] Among them, σ(P i,l ) represents the frequency variance of the lth dimension in the i-th stage, N i,l represents the number of data pairs in the data set of the lth dimension in the i-th stage, d i,l,n It represents the distance between two data in the nth data pair in the data set of the lth dimension in the i-th stage. represents the maximum value of the distance between two data in all data pairs in the data set of the lth dimension in the i-th stage, f i,l (j) represents the data of the jth student in the i-th stage and the l-th dimension, represents the mean of the data of all students in the lth dimension of the i-th stage, max(Δf i,l) represents the maximum absolute value of the difference between the data of all students in the lth dimension of the i-th stage and the mean, || represents the absolute value, and exp[] represents the exponential function with a natural constant as the base;

[0016] Obtain the baseline value of attention for each student at each stage and in each dimension.

[0017] Furthermore, the correction weight value of each dimension of each student at each stage is obtained in the following specific method:

[0018] For the data of the jth student at different stages of the lth dimension, the horizontal axis is the stage and the vertical axis is the data value. The stage data change curve of the dimension of the student is constructed. The slope of each data point in the stage data change curve except the first data point is calculated. The correction weight value ε of the jth student in the lth dimension in the i stage is i，l The calculation method of (j) is:

[0019]

[0020]

[0021] in, represents the slope change parameter of the jth student in the i-th stage, the l-th dimension, f i，l (j) represents the data of the jth student in the i-th stage, the l-th dimension, T 1，l (j) represents the data of the jth student in the lth dimension in the first stage, (j) represents the mean of the absolute value of the difference between the data of the j-th student in the i-th stage and all adjacent stages before the l-th dimension, || represents the absolute value, and exp() represents the exponential function with a natural constant as the base; M i，l (j) represents the number of slopes of the jth student in the i-th stage and the previous stage in the stage data change curve of the first dimension; M1 i，l (j) represents the number of positive slopes in the i-th stage and the previous stages in the stage data change curve of the j-th student in the l-th dimension; M2 i，l (j) represents the number of negative slopes in the i-th stage and the previous stages in the stage data change curve of the j-th student in the first dimension;

[0022] Obtain the corrected weight value for each dimension of each student at each stage.

[0023] Furthermore, the method of obtaining the attention level of each dimension of each student at each stage according to the corrected weight value and the attention level benchmark value includes the following specific methods:

[0024] For the corrected weight value of the first dimension of the j-th student in the i-th stage, add 1 to the corrected weight value, and multiply it by the baseline value of the attention of the j-th student in the first dimension in the i-th stage, and use it as the attention of the j-th student in the first dimension in the i-th stage;

[0025] Get the attention of each student at each stage and in each dimension.

[0026] Furthermore, the specific method for obtaining the difference distance between any two students at the same stage is as follows:

[0027] Take any student as the current student, perform softmax normalization on the attention of all dimensions of the current student at any stage, and record the result as the attention weight of each dimension of the current student at that stage;

[0028] Obtain the data of all dimensions of the current student and any other student except the current student at this stage, obtain the absolute value of the difference between the data of the same dimension of the two students at this stage, obtain the absolute value of the difference of each dimension, perform linear normalization on the absolute value of the difference, and record the result as the degree of difference between the student and the current student at this stage in each dimension;

[0029] Take the weighted sum of all the difference degrees according to the current student's attention weight of the corresponding dimension, and record the sum as the difference coefficient between the student and the current student at this stage; obtain the difference coefficient between each other student and the current student at this stage; obtain the difference coefficient between any two students at this stage;

[0030] There are two difference coefficients between any two students at this stage. The maximum value of the two difference coefficients is taken as the difference distance between the two students at this stage, and the difference distance between any two students at this stage is obtained;

[0031] Get the difference distance between any two students at each stage.

[0032] Furthermore, the clustering based on the difference distance is performed to obtain the cluster to which each student belongs in each stage, including the following specific methods:

[0033] For any stage, DBSCAN clustering is performed on the students in that stage according to the difference distance. The clustering distance metric is the difference distance between two students in that stage. All students in that stage are divided into several clusters, and the cluster to which each student belongs in that stage is obtained; the cluster to which each student belongs in each stage is obtained.

[0034] Furthermore, the degree of deviation of each dimension of each student at each stage is obtained in the following specific method:

[0035] For any cluster, each student in the cluster is treated as a data point, a distance neighborhood size is preset, and the local distance neighborhood of each data point is obtained according to the distance neighborhood size. The density of the local distance neighborhood of each data point is obtained, and the density of the local distance neighborhood of each student in the cluster is obtained.

[0036] The cluster is classified according to the density of each student's local distance neighborhood, and two categories are obtained using the maximum inter-class variance method. The category with the smallest average density of the local distance neighborhood between the two categories is used as the standard category for the cluster. For any dimension, the minimum value of the data values ​​of all students in the standard category in that dimension is used as the adaptive threshold of that dimension in the cluster, and the adaptive threshold of that dimension for each student in the cluster at the corresponding stage is obtained.

[0037] Obtain the density of each student's local distance neighborhood in each cluster, obtain the adaptive threshold of each dimension in each cluster, and obtain the adaptive threshold of each dimension in each student's cluster at each stage;

[0038] For the data of any student in any dimension at any stage, the difference between the adaptive threshold of the dimension in the cluster to which the student belongs at that stage is subtracted from the data to obtain the degree of deviation of the student in that dimension at that stage; the degree of deviation of each dimension of each student at each stage is obtained.

[0039] Furthermore, the specific method for obtaining the deviation value of each dimension for each student is as follows:

[0040] The latest stage of obtaining data of multiple dimensions is taken as the current stage, expressed as the p-th stage, and the accumulated weight value δ of the l-th dimension of the j-th student in the current stage is p,l The calculation method of (j) is:

[0041]

[0042]

[0043] Where y(s) represents the output value of the cumulative mapping function for s, T s Indicates the time corresponding to the sth stage, T s-1 Indicates the time corresponding to the s-1th stage, T max Indicates the maximum time corresponding to the stage before the current stage, ω s,l (j) represents the degree of deviation of the jth student in the sth stage of the lth dimension, ω s-1,l (j) represents the degree of deviation of the lth dimension of the jth student in the s-1th stage, and || represents the absolute value;

[0044] The sum of the accumulated weight values ​​plus 1 is multiplied by the deviation degree of the j-th student in the l-th dimension at the current stage, and the product is used as the deviation parameter of the j-th student in the l-th dimension at the current stage. The deviation parameters of each dimension of the j-th student in the current stage are obtained, and all the deviation parameters are linearly normalized. The result is recorded as the deviation value of each dimension of the j-th student;

[0045] Get the accumulated weight value of each dimension of each student in the current stage, get the deviation parameter of each dimension of each student in the current stage, and get the deviation value of each dimension of each student.

[0046] Furthermore, the method of performing adaptive PCA dimensionality reduction on each student and obtaining a user portrait includes the following specific methods:

[0047] For any dimension of any student, arrange the data of all stages under that dimension in the chronological order of the stages to obtain a vector, which is recorded as the dimension vector of that dimension of the student; obtain the dimension vector of each dimension of the student, use the deviation value of each dimension of the student as the eigenvalue of the corresponding dimension vector, perform PCA dimensionality reduction on all dimension vectors, and obtain the user profile of the student;

[0048] Obtain the dimension vector for each dimension of each student, and perform adaptive PCA dimensionality reduction based on the corresponding deviation value to obtain the user portrait of each student.

[0049] In a second aspect, another embodiment of the present invention provides a comprehensive management system for collaborative education information data between a home, school, and community, the system comprising:

[0050] Student multi-dimensional data collection module, used to collect data of each dimension of each stage for several students;

[0051] The multidimensional data processing and analysis module is used to obtain the attention benchmark value of each dimension at each stage for each student based on the distribution of all data under the same dimension for all students at the same stage; obtain the corrected weight value of each dimension at each stage for each student based on the changes in data under the same dimension at different stages for the same student; and obtain the attention degree of each dimension at each stage for each student based on the corrected weight value and the attention benchmark value;

[0052] Based on the differences in data of different students at the same stage and the same dimension, as well as the attention of each student at each stage and each dimension, the difference distance between any two students at the same stage is obtained; clustering is performed based on the difference distance to obtain the cluster to which each student belongs at each stage;

[0053] Based on the adaptive threshold of each dimension of each cluster and the data of each dimension of each student at each stage, the deviation degree of each dimension of each student at each stage is obtained; based on the deviation degree of the same dimension of each student at different stages, the deviation value of each dimension of each student is obtained;

[0054] The user portrait acquisition management module is used to perform adaptive PCA dimensionality reduction on each student based on the dimension vector and deviation value of each dimension of each student and obtain the user portrait, which is stored in the comprehensive information data management of collaborative education between home, school and community.

[0055] The beneficial effects of the present invention are as follows: the present invention analyzes the data distribution characteristics of each dimension of students in the same stage, thereby obtaining the attention level of each dimension of each stage of the students, clusters the students in different stages according to the attention level, and obtains student groups in different stages that have strong similarities with the data of the current students; obtains the deviation change of the students' multidimensional data in the student group, and at the same time analyzes the change of the students' multidimensional information in different stages of the students, and judges the degree of dimensionality reduction target of different dimensions of the students' multidimensional data in the process of adaptive PCA dimensionality reduction, that is, performs adaptive PCA dimensionality reduction according to the deviation value. It avoids the problem of information loss in the process of traditional PCA dimensionality reduction, and ensures the stability of the dimensions that students focus on over time, so that the results in the clustering process are more accurate, and then the results are more accurate when calculating the deviation change of the current students' data in different dimensions, and can generate more accurate user portraits to assist parents and teachers in educating students. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0057] Figure 1 A schematic flow chart of a method for comprehensive management of educational information data in a collaborative manner between home, school and community provided by one embodiment of the present invention;

[0058] Figure 2 This is a structural block diagram of a comprehensive management system for collaborative education information data between home, school and community provided by another embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] See also Figure 1 , which shows a flow chart of a method for comprehensive management of home-school-community collaborative education information data provided by one embodiment of the present invention, the method comprising the following steps:

[0061] Step S001: Collect data of each dimension of each stage for several students.

[0062] The purpose of this embodiment is to comprehensively manage the information data of students under the collaborative education of home, school and community. Therefore, it is necessary to obtain data of students in multiple dimensions under home, school and community to obtain information data and comprehensively manage it. This embodiment first needs to obtain the name, gender and age of the students as the basic information of the students. The basic information is directly stored and does not participate in the analysis of subsequent dimensional data. As for the data of multiple dimensions, its sources include the academic affairs system, online learning platform, family questionnaire, community activity record system and health and exercise system. This embodiment collects students' test scores, attendance records and homework submission completion through the academic affairs system, collects students' course completion and test scores through the online learning platform, collects parents' observations and feedback on students through the family questionnaire, and collects community activity records through the community activity record. The recording system collects the number and time of extracurricular activities, club activities and volunteer services participated by students; the health and exercise system collects various data value indicators in students' exercise records and health examination reports; the various collected data are respectively used as data of a dimension, wherein the collection time interval of different dimensions is set to one month in this embodiment, and one month is used as a stage, that is, the data of each dimension is collected every other month. At the same time, due to the difference in the dimensions of data of different dimensions, this embodiment uses numerical standardization processing to make the value range of the data of each dimension range from [0,1]. The numerical standardization processing is a well-known technology and will not be repeated in this embodiment; the data of each dimension of each student in each stage are obtained to form a data set for subsequent analysis.

[0063] At this point, data on each dimension of each student at each stage has been obtained.

[0064] Step S002: Based on the distribution of all data under the same dimension for all students at the same stage, obtain the attention baseline value for each dimension at each stage for each student; based on the changes in data under the same dimension at different stages for the same student, obtain the corrected weight value for each dimension at each stage for each student; and obtain the attention degree for each dimension at each stage for each student based on the corrected weight value and the attention baseline value.

[0065] It should be noted that by obtaining students' multidimensional data and obtaining changes in students' data during the continuous learning process, the user portrait of the students is updated. In the process of updating the user portrait, the collected multidimensional data needs to be subjected to data dimensionality reduction analysis. However, the dimensionality reduction data obtained by traditional dimensionality reduction methods such as the PCA algorithm will have the problem of losing information, and because there are differences in the dimensions that students focus on over time, the deviation changes in the student's dimensional data are obtained by analyzing the student groups that have a strong similarity with the data at different stages, and analyzing the changes in the student's multidimensional information at different stages. To judge the dimensionality reduction target degree of different dimensions of the student's multidimensional data in the process of adaptive PCA dimensionality reduction, that is, the eigenvalue, and then perform adaptive PCA dimensionality reduction.

[0066] It should be further explained that in the process of adaptive PCA dimensionality reduction of students' multidimensional data, due to the different stages of students, such as the grade of students, different students at different stages focus on different dimensions. For example, students in lower grades care more about extracurricular activities, interest activities, etc., while students in higher grades care more about learning and grades. At the same time, as the stages change, the dimensions that current students care about may change with different stages. For example, if the English grades are very good in the lower grades, they will not care much about the English grades, but as they reach higher grades, their English grades will deteriorate, and they will care more and more about the English grades. Correspondingly, with different stages, they will never care about the English grades but care more about the English grades. Therefore, when analyzing students in PCA dimensionality reduction, it is necessary to obtain the clusters in which the students are located, and in the process of clustering, it is necessary to simultaneously analyze and obtain the students' attention to each dimension in different stages.

[0067] Specifically, taking the first dimension of the i-th stage as an example, obtain the data of all students in this dimension in this stage as the data set of this dimension in this stage, obtain the frequency of occurrence of each data in the data set, calculate the variance according to the frequency of occurrence, and record the result as the frequency variance of this dimension in this stage; calculate the absolute value of the difference between any two data in the data set, take the absolute value of the difference as the distance between the two data, start extracting from the minimum value of all distances, and obtain several data pairs without repetition, that is, first extract the two data corresponding to the minimum distance as a data pair, and then obtain the two data corresponding to the minimum distance from the remaining data as a data pair, until all data in the data set are divided into corresponding data pairs. It should be noted that if the number of data in the data set is an odd number, the last remaining data will not participate in the division of the data pairs; then the attention baseline value α of the j-th student in the l-th dimension of the i-th stage is i，l The calculation method of (j) is:

[0068]

[0069] Among them, σ(P i，l ) represents the frequency variance of the lth dimension in the i-th stage, N i，l represents the number of data pairs in the data set of the first dimension in the i-th stage, d i，l，n Indicates the distance between two data in the nth data pair in the data set of the first dimension in the i-th stage, represents the maximum value of the distance between two data in all data pairs in the data set of the first dimension of the i-th stage, f i，l (j) represents the data of the first dimension of the jth student in the i-th stage, represents the mean of the data of all students in the lth dimension of the i-th stage, max(Δf i，l ) represents the maximum absolute value of the difference between the data of all students in the lth dimension in the i-th stage and the mean, || represents the absolute value, exp[] represents an exponential function with a natural constant as the base. This embodiment uses the exp[-x] model to present the inverse proportional relationship and normalization processing, where x is the input of the model. The implementer can set the inverse proportional function and normalization function according to the actual situation;

[0070] First, the clustering characteristics of the data set are quantified by the frequency variance and the distance difference of data pairs in the data set. If the frequency variance is smaller, it means that the occurrence frequency of various data is closer, the data distribution is more discrete, and the corresponding clustering characteristics are smaller; at the same time, the distance of data pairs in the data set can also reflect the clustering characteristics. The smaller the distance of data pairs, the more clustered the data distribution in the data set, and the greater the clustering characteristics; and for the data of the student in the data set, the greater the clustering characteristics, if the difference with the data mean is greater, the relative difference characteristics are greater, and thus the attention baseline value is greater; according to the above method, the attention baseline value of each dimension of each student at each stage is obtained.

[0071] Furthermore, for the data of the jth student at different stages of the first dimension, with the horizontal axis as the stage and the vertical axis as the data value, a stage data change curve of the student in this dimension is constructed. The slope of each data point in the stage data change curve except the first data point is calculated. Then the correction weight value ε of the jth student in the first dimension in the i stage is i，l The calculation method of (j) is:

[0072]

[0073]

[0074] in, represents the slope change parameter of the jth student in the first dimension of the i-th stage, fi,l(j) represents the data of the jth student in the first dimension of the i-th stage, f 1，l (j) represents the data of the jth student in the first dimension of the first stage, (j) represents the mean of the absolute value of the difference between the data of the j-th student in the i-th stage and all adjacent stages before the first dimension, || represents the absolute value, exp() represents the exponential function with a natural constant as the base. This embodiment uses the exp(-x) model to present the inverse proportional relationship and normalization processing, where x is the input of the model. The implementer can set the inverse proportional function and normalization function according to the actual situation; M i，l (j) represents the number of slopes of the jth student in the i-th stage and the previous stage in the stage data change curve of the first dimension; M1 i，l (j) represents the number of positive slopes in the i-th stage and the previous stages in the stage data change curve of the j-th student in the first dimension; M2 i，l (j) represents the number of negative slopes in the i-th stage and the previous stages in the stage data change curve of the j-th student in the l-th dimension;

[0075] The larger the slope change parameter is, the more slopes of the same sign are found in the stage data change curve of the dimension for this student. If the stage data change curve as a whole shows an increasing or decreasing trend, the difference between the data of the i-th stage and the first stage should be considered more to quantify the correction weight value. The greater the difference between the two data, the more it is necessary to adjust the attention baseline value, and thus increase the attention baseline value. The smaller the slope change parameter is, the smaller the difference in the number of slopes of different signs in the stage data change curve is, and the data of adjacent stages change unevenly. In this case, the correction weight value needs to consider the changes in all adjacent stage data. The greater the mean of the difference between the adjacent stage data, the greater the correction weight value should be, and thus increase the attention baseline value.

[0076] Furthermore, after obtaining the corrected weight value of the jth student in the lth dimension in the i-th stage, the sum of the corrected weight value plus 1 and the attention baseline value of the jth student in the l-th dimension in the i-th stage is multiplied as the attention of the jth student in the l-th dimension in the i-th stage; the corrected weight value of each dimension in each stage of each student is obtained according to the above method, and the attention of each dimension in each stage of each student is obtained.

[0077] So far, we have gained attention from every student at every stage and in every dimension.

[0078] Step S003: Based on the data differences of different students in the same stage and the same dimension, and the attention of each student in each dimension in each stage, the difference distance between any two students in the same stage is obtained; clustering is performed based on the difference distance to obtain the cluster to which each student belongs in each stage.

[0079] It should be noted that after obtaining the attention of each student in each dimension at each stage, it is necessary to perform cluster analysis on each student at different stages, where the cluster distance measurement is quantified according to the difference in data of each dimension of different students at the same stage. At the same time, since the attention of each student in each dimension at each stage is different, the difference in data of each dimension can be weighted averaged according to the attention, and then the difference distance between any two students at the same stage can be obtained, and cluster analysis can be performed based on the difference distance.

[0080] Specifically, take any student as the current student, perform softmax normalization on the attention of all dimensions of the current student at any stage, and record the result as the attention weight of each dimension of the current student at this stage; obtain the data of all dimensions of the current student and any student other than the current student at this stage, obtain the absolute value of the difference between the data of the same dimension of the two students at this stage, obtain the absolute value of the difference of each dimension, perform linear normalization on the absolute value of the difference, and record the result as the degree of difference between the student and the current student at this stage in each dimension; perform weighted summation of all difference degrees according to the attention weight of the current student in the corresponding dimension, and record the sum as the difference coefficient between the student and the current student at this stage; obtain the difference coefficient between each other student and the current student at this stage according to the above method, and obtain the difference coefficient between any two students at this stage; since the acquisition of the difference coefficient is related to the student serving as the benchmark, that is, related to the current student, there are two difference coefficients for any two students at this stage, and the maximum value of the two difference coefficients is taken as the difference distance between the two students at this stage, and the difference distance between any two students at this stage is obtained.

[0081] Furthermore, DBSCAN clustering is performed on the students in this stage according to the difference distance. The distance metric of the clustering is the difference distance between the two students in this stage. Then all the students in this stage can be divided into several clusters, and the cluster to which each student belongs in this stage is obtained. The difference distance between any two students in each stage is obtained according to the above method, and the students in each stage are clustered according to the difference distance to obtain the cluster to which each student belongs in each stage.

[0082] At this point, based on the data and attention of each student in each dimension at each stage, the cluster to which each student belongs at each stage is obtained.

[0083] Step S004: Based on the adaptive threshold of each dimension of each cluster and the data of each dimension of each student at each stage, obtain the degree of deviation of each dimension of each student at each stage; based on the degree of deviation of the same dimension of each student at different stages, obtain the deviation value of each dimension of each student.

[0084] It should be noted that after obtaining the cluster to which each student belongs in each stage, the students in the cluster have certain similarities in the corresponding stage. In order to quantify the degree of deviation of each student in the corresponding stage according to the cluster, it is necessary to obtain the threshold of each cluster for quantifying the degree of deviation, wherein the threshold is obtained by quantifying the density of the local distance neighborhood of the students in the cluster. Through analysis similar to the LOF algorithm, the density of the local distance neighborhood of each student is obtained, and the density is classified according to the maximum inter-class variance method to determine the maximum difference within the cluster. By obtaining a category with the smallest density in this classification, this category reflects that the density of the student's local distance neighborhood is small, that is, the distribution within the cluster has begun to be sparse, then the adaptive threshold of each dimension is obtained according to the category, and the degree of deviation of the data of each dimension of each student at each stage is obtained according to the adaptive threshold.

[0085] It should be further explained that after obtaining the degree of deviation, it is necessary to analyze the cumulative changes in the degree of deviation. According to the degree of deviation of the same dimension of the student at all stages, the cumulative weight value of each dimension in the current stage of the student is obtained, and the degree of deviation in the current stage is adjusted to obtain the deviation value of each dimension in the current stage as the characteristic value. This reflects the changes in the students' emphasis on dimensions at different stages, as well as the changes in the emphasis on dimensions brought about by the changing characteristics of the data of the corresponding dimensions.

[0086] Specifically, for any cluster containing several students, each student is taken as a data point, and a distance neighborhood size is preset. The distance neighborhood size in this embodiment is described as 6, that is, the 6th distance neighborhood size is obtained for each data point to form the local distance neighborhood of each data point. The local distance neighborhood is obtained similar to the kth distance neighborhood in the LOF algorithm. The LOF algorithm is a well-known technology and will not be described in detail in this embodiment. The distance of the data points is the difference distance between the students in the cluster; after obtaining the local distance neighborhood of each data point, the density of the local distance neighborhood of each data point is obtained. The density calculation is also a well-known technology and will not be described in detail in this embodiment; the density of the local distance neighborhood of each student in the cluster is obtained.

[0087] Furthermore, the cluster is classified into two categories according to the density of the local distance neighborhood of each student. The maximum inter-class variance method is used to obtain the classification with the greatest difference, and two categories are obtained. That is, at this time, the difference in the density of the local distance neighborhood in the two categories is the largest. The maximum inter-class variance method is a well-known technology and will not be repeated in this embodiment. The category with the smallest average value of the density of the local distance neighborhood in the two categories is used as the standard category of the cluster. For any dimension, the minimum value of the data values ​​of all students in the standard category in this dimension is used as the adaptive threshold of this dimension in the cluster. Since the cluster is based on the stage, the adaptive threshold of this dimension of each student in the cluster in the corresponding stage can be obtained. According to the above method, the density of the local distance neighborhood of each student in each cluster is obtained, and then the adaptive threshold of each dimension in each cluster is obtained, and the adaptive threshold of each dimension in the cluster to which each student belongs in each stage is obtained.

[0088] Furthermore, for any student’s data of any dimension in any stage, the difference between the adaptive threshold of the dimension in the cluster to which the student belongs in that stage is subtracted from the data is taken as the deviation degree of the student in that dimension in that stage; the deviation degree of each dimension in each stage of each student is obtained according to the above method; the stage in which the data of multiple dimensions are obtained most recently is taken as the current stage, denoted as the pth stage, and the accumulated weight value δ of the lth dimension of the jth student in the current stage is p，l The calculation method of (j) is:

[0089]

[0090]

[0091] Where y(s) represents the output value of the cumulative mapping function for s. Its expression means that if the deviation degree of adjacent stages decreases, the corresponding student is closer to other students in the cluster in the adjacent stage, that is, the similarity becomes greater, and the corresponding impact of the deviation degree on the subsequent stage will be smaller, and the corresponding deviation degree difference should be reduced; while if the deviation degree of adjacent stages increases, the corresponding student is farther away from other students in the cluster in the adjacent stage, and the impact of the deviation degree on the subsequent stage will be greater, and the corresponding deviation degree difference should be increased.

[0092] Among them, T s Indicates the time corresponding to the sth stage, T s-1 Indicates the time corresponding to the s-1th stage, T max Indicates the maximum time corresponding to the current stage, that is, the stage before the pth stage, that is, T max =T p-1 ;ω s，l (j) represents the degree of deviation of the first dimension of the jth student in the sth stage, ωs-1，l (j) represents the degree of deviation of the jth student in the s-1th stage of the lth dimension, and || represents the absolute value. By quantifying the difference in the degree of deviation of all adjacent stages before the current stage, and taking the time difference between the corresponding two stages and the current stage as the weight, the larger the time difference, the smaller the confidence should be, and the smaller the weight should be, then At the same time, the degree of deviation is accumulated, and combined with the cumulative mapping function, the cumulative weight value of the first dimension of the j-th student in the current stage is finally obtained.

[0093] Furthermore, the sum of the accumulated weight values ​​added by 1 is multiplied by the deviation degree of the first dimension of the j-th student in the current stage, and the product is used as the deviation parameter of the first dimension of the j-th student in the current stage. The deviation parameter of each dimension of the j-th student in the current stage is obtained, and all deviation parameters are linearly normalized. The result is recorded as the deviation value of each dimension of the j-th student; the accumulated weight value of each dimension of each student in the current stage is obtained according to the above method, and the deviation parameter of each dimension of each student in the current stage is obtained, and then the deviation value of each dimension of each student is obtained.

[0094] At this point, the deviation value of each dimension for each student has been obtained. By analyzing the degree of deviation of each dimension in the cluster to which each student belongs at different stages and combining the cumulative effect of the degree of deviation, we can finally obtain a deviation value that can reflect the changes in the emphasis of the dimension brought about by the changes in each student's stage, as well as the changes in the emphasis of the dimension brought about by the changing characteristics of the dimension data itself.

[0095] Step S005: Based on the dimension vector and deviation value of each dimension of each student, perform adaptive PCA dimensionality reduction on each student and obtain a user portrait to complete the comprehensive management of information data for collaborative education between home, school and community.

[0096] For any dimension of any student, the data of all stages under this dimension of the student are arranged in chronological order of the stages to obtain a vector, which is recorded as the dimension vector of this dimension of the student; the dimension vector of each dimension of the student is obtained according to the above method, the deviation value of each dimension of the student is used as the eigenvalue of the corresponding dimension vector, and PCA dimensionality reduction is performed on all dimensional vectors to obtain the user portrait of the student; the dimension vector of each dimension of each student is obtained according to the above method, and adaptive PCA dimensionality reduction is performed according to the corresponding deviation value to obtain the user portrait of each student, wherein PCA dimensionality reduction to obtain the user portrait is a well-known technology and will not be repeated in this embodiment.

[0097] Furthermore, since each student is acquiring data of a new stage in each dimension, and the dimension vectors and deviation values ​​are all acquired up to the current stage, the corresponding dimension vectors and deviation values ​​will also change as the data of the new stage is acquired, that is, the dimension vectors and deviation values ​​are updated with the stage, and the student's user portrait can be updated; the user portrait of each student is stored in the home-school-community collaborative education information data comprehensive management platform, and personalized labels are displayed to assist teachers and parents in educating students; at the same time, after the student's user portrait is updated, the user portrait in the platform is updated.

[0098] At this point, the comprehensive management of information data on students' collaborative education between home, school and community has been completed.

[0099] See also Figure 2 , which shows a structural block diagram of a comprehensive management system for collaborative education information data between a home, school and community provided by another embodiment of the present invention, the system comprising:

[0100] The student multi-dimensional data collection module 101 collects data of each dimension of each stage for several students.

[0101] Multidimensional data processing and analysis module 102:

[0102] (1) According to the distribution of all data of all students in the same stage and the same dimension, obtain the attention benchmark value of each student in each stage and each dimension; according to the changes of data of the same dimension in different stages of the same student, obtain the correction weight value of each dimension in each stage of each student; according to the correction weight value and the attention benchmark value, obtain the attention of each student in each dimension in each stage;

[0103] (2) Based on the differences in data of different students at the same stage and the same dimension, as well as the attention of each student at each stage and each dimension, the difference distance between any two students at the same stage is obtained; clustering is performed based on the difference distance to obtain the cluster to which each student belongs at each stage;

[0104] (3) Based on the adaptive threshold of each dimension of each cluster and the data of each dimension of each student at each stage, the degree of deviation of each dimension of each student at each stage is obtained; based on the degree of deviation of the same dimension of each student at different stages, the deviation value of each dimension of each student is obtained.

[0105] The user portrait acquisition management module 103 performs adaptive PCA dimensionality reduction on each student based on the dimension vector and deviation value of each dimension of each student and obtains a user portrait, thereby completing the comprehensive management of information data for collaborative education between home, school and community.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A comprehensive management method for collaborative education information data between home, school and community, characterized by: The method comprises the following steps: Collect data from several students in each dimension at each stage; According to the distribution of all data under the same dimension of the same stage for all students, obtain the attention benchmark value of each dimension of each student at each stage, the attention benchmark value is determined according to the frequency variance of each dimension, the number of data pairs in the data set of each dimension, the distance between two data in all data pairs in the data set and the data of each dimension. The data set is the data of all students in each dimension at each stage, the frequency variance is determined according to the occurrence frequency of each data in the data set, and the frequency variance and the distance are used to reflect the aggregation characteristics of the data set; according to the change of data under the same dimension at different stages of the same student, obtain the correction weight value of each dimension of each student at each stage, the correction weight value is determined according to the slope change parameter of each dimension, the data of each dimension and the stage data change curve of each dimension, the horizontal axis of the stage data change curve is the stage, and the vertical axis is the data value; according to the correction weight value and the attention benchmark value, obtain the attention of each dimension of each stage for each student; Based on the differences in data of different students at the same stage and the same dimension, as well as the attention of each student at each stage and each dimension, the difference distance between any two students at the same stage is obtained; clustering is performed based on the difference distance to obtain the cluster to which each student belongs at each stage; According to the adaptive threshold of each dimension of each cluster and the data of each dimension of each student at each stage, the degree of deviation of each dimension of each student at each stage is obtained; according to the degree of deviation of the same dimension of each student at different stages, the deviation value of each dimension of each student is obtained, wherein the deviation value of each dimension is determined according to the deviation parameter of each dimension, and the deviation parameter is determined according to the accumulated weight value of each dimension and the degree of deviation of each dimension, and the accumulated weight value is determined according to the time corresponding to each stage and the degree of deviation of the same dimension at different stages; According to the dimension vector and deviation value of each dimension of each student, adaptive PCA dimensionality reduction is performed on each student to obtain a user portrait, which is stored in the comprehensive information data management of collaborative education between home, school and community.

2. A comprehensive management method for collaborative education information data between home, school and community according to claim 1, characterized in that: The specific method for obtaining the attention benchmark value of each dimension of each student at each stage is as follows: For the Phase I Dimension, obtain the data of all students in this dimension at this stage as the data set of this dimension at this stage, obtain the frequency of occurrence of each data in the data set, calculate the variance based on the frequency of occurrence, and record the result as the frequency variance of this dimension at this stage; Calculate the absolute value of the difference between any two data in the data set, take the absolute value of the difference as the distance between the two data, start extracting from the minimum value of all distances, and obtain several data pairs without duplication; Students in the Phase I The attention benchmark value of each dimension The calculation method is: in, Indicates the Phase I The frequency variance of the dimension, Indicates the Phase I The number of data pairs in the dataset with dimensions, Indicates the Phase I The distance between two data in the nth data pair in a data set of dimensions, Indicates the Phase I The maximum value of the distance between two data in all data pairs in the data set of dimensions, Indicates the Students in the Phase I Dimensional data, Indicates the Phase I The mean of the data of all students in the dimension, Indicates the Phase I The maximum absolute value of the difference between the data of all students in a dimension and the mean, Indicates the absolute value. represents an exponential function with a natural constant as its base; Obtain the baseline value of attention for each student at each stage and in each dimension.

3. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method for obtaining the correction weight value of each dimension of each student at each stage is as follows: For the Students in the The data of different stages of each dimension are taken as the horizontal axis, and the vertical axis is the data value. The stage data change curve of the dimension of the student is constructed. The slope of each data point except the first data point in the stage data change curve is calculated. Students in the Phase I Correction weight value of each dimension The calculation method is: in, Indicates the Students in the Phase I The slope change parameter of each dimension, Indicates the Students in the Phase I Dimensional data, Indicates the Students in the first stage Dimensional data, Indicates the Students in the Dimension The mean of the absolute value of the difference between the data of the stage and all adjacent stages before the stage, Indicates the absolute value. represents an exponential function with a natural constant as its base; Indicates the Students in the In the stage data change curve of the dimensions, the number of slopes in the stage and the preceding stages; Indicates the Students in the In the stage data change curve of the dimensions, The number of slopes with positive slope values ​​in the stage and the previous stage; Indicates the Students in the In the stage data change curve of the dimensions, The number of slopes with negative slope values ​​in the stage and the previous stage; Obtain the corrected weight value for each dimension of each student at each stage.

4. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method of obtaining the attention level of each dimension of each student at each stage according to the corrected weight value and the attention level benchmark value includes: For the Students in the Phase I The correction weight value of the dimension is added by 1, and the sum is obtained by adding 1 to the correction weight value. Students in the Phase I The product of the attention baseline values ​​of the dimensions is used as the Students in the Phase I The degree of attention in each dimension; Get the attention of each student at each stage and in each dimension.

5. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method for obtaining the difference distance between any two students at the same stage is as follows: Take any student as the current student, perform softmax normalization on the attention of all dimensions of the current student at any stage, and record the result as the attention weight of each dimension of the current student at that stage; Obtain the data of all dimensions of the current student and any other student except the current student at this stage, obtain the absolute value of the difference between the data of the same dimension of the two students at this stage, obtain the absolute value of the difference of each dimension, perform linear normalization on the absolute value of the difference, and record the result as the degree of difference between the student and the current student at this stage in each dimension; Take the weighted sum of all the difference degrees according to the current student's attention weight of the corresponding dimension, and record the sum as the difference coefficient between the student and the current student at this stage; obtain the difference coefficient between each other student and the current student at this stage; obtain the difference coefficient between any two students at this stage; There are two difference coefficients between any two students at this stage. The maximum value of the two difference coefficients is taken as the difference distance between the two students at this stage, and the difference distance between any two students at this stage is obtained; Get the difference distance between any two students at each stage.

6. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method of clustering based on difference distance to obtain the cluster to which each student belongs in each stage is as follows: For any stage, DBSCAN clustering is performed on the students in that stage according to the difference distance. The clustering distance metric is the difference distance between two students in that stage. All students in that stage are divided into several clusters, and the cluster to which each student belongs in that stage is obtained; the cluster to which each student belongs in each stage is obtained.

7. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method for obtaining the degree of deviation of each dimension of each student at each stage is as follows: For any cluster, each student in the cluster is treated as a data point, a distance neighborhood size is preset, and the local distance neighborhood of each data point is obtained according to the distance neighborhood size. The density of the local distance neighborhood of each data point is obtained, and the density of the local distance neighborhood of each student in the cluster is obtained. The cluster is classified according to the density of each student's local distance neighborhood, and two categories are obtained using the maximum inter-class variance method. The category with the smallest average density of the local distance neighborhood between the two categories is used as the standard category for the cluster. For any dimension, the minimum value of the data values ​​of all students in the standard category in that dimension is used as the adaptive threshold of that dimension in the cluster, and the adaptive threshold of that dimension for each student in the cluster at the corresponding stage is obtained. Obtain the density of each student's local distance neighborhood in each cluster, obtain the adaptive threshold of each dimension in each cluster, and obtain the adaptive threshold of each dimension in each student's cluster at each stage; For the data of any student in any dimension at any stage, the difference between the adaptive threshold of the dimension in the cluster to which the student belongs at that stage is subtracted from the data to obtain the degree of deviation of the student in that dimension at that stage; the degree of deviation of each dimension of each student at each stage is obtained.

8. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method for obtaining the deviation value of each dimension of each student is as follows: The latest stage of acquiring data of multiple dimensions is taken as the current stage, which is expressed as the pth stage, Students in the current stage The cumulative weight value of each dimension The calculation method is: in, Represents the output value of the cumulative mapping function for S, represents the time corresponding to the Sth stage, Indicates the The time corresponding to each stage, Indicates the maximum time corresponding to the stage before the current stage, Indicates the Students Sth stage The degree of deviation in each dimension, Indicates the Student No. Phase I The degree of deviation in each dimension, Indicates finding the absolute value; The sum of the accumulated weight value plus 1 is Students in the current stage The product of the deviation degree of the dimensions is used as the Students in the current stage The deviation parameter of the dimension is obtained The deviation parameters of each dimension of the current stage of each student are linearly normalized for all deviation parameters, and the result is recorded as Deviation value of each dimension for each student; Get the accumulated weight value of each dimension of each student in the current stage, get the deviation parameter of each dimension of each student in the current stage, and get the deviation value of each dimension of each student.

9. The method for comprehensive management of home-school-community collaborative education information data according to claim 1, characterized in that: The specific method of performing adaptive PCA dimensionality reduction on each student and obtaining a user profile includes: For any dimension of any student, arrange the data of all stages under that dimension in the chronological order of the stages to obtain a vector, which is recorded as the dimension vector of that dimension of the student; obtain the dimension vector of each dimension of the student, use the deviation value of each dimension of the student as the eigenvalue of the corresponding dimension vector, perform PCA dimensionality reduction on all dimension vectors, and obtain the user profile of the student; Obtain the dimension vector for each dimension of each student, and perform adaptive PCA dimensionality reduction based on the corresponding deviation value to obtain the user portrait of each student.

10. A comprehensive management system for collaborative education information data between home, school and community, characterized by: The system includes: Student multi-dimensional data collection module, used to collect data of each dimension of each stage for several students; A multidimensional data processing and analysis module is used to obtain a baseline value of attention for each dimension of each student at each stage based on the distribution of all data under the same dimension of the same stage for all students. The baseline value of attention is determined based on the frequency variance of each dimension, the number of data pairs in the data set of each dimension, the distance between two data in all data pairs in the data set, and the data of each dimension. The data set is the data of all students in each dimension at each stage. The frequency variance is determined based on the frequency of occurrence of each data in the data set. The frequency variance and the distance are used to reflect the aggregation characteristics of the data set. According to the changes in the data of the same dimension at different stages of the same student, a correction weight value of each dimension at each stage of each student is obtained. The correction weight value is determined based on the slope change parameter of each dimension, the data of each dimension, and the stage data change curve of each dimension. The horizontal axis of the stage data change curve is the stage, and the vertical axis is the data value. According to the correction weight value and the attention baseline value, the attention of each dimension of each student at each stage is obtained. Based on the differences in data of different students at the same stage and the same dimension, as well as the attention of each student at each stage and each dimension, the difference distance between any two students at the same stage is obtained; clustering is performed based on the difference distance to obtain the cluster to which each student belongs at each stage; According to the adaptive threshold of each dimension of each cluster and the data of each dimension of each student at each stage, the degree of deviation of each dimension of each student at each stage is obtained; according to the degree of deviation of the same dimension of each student at different stages, the deviation value of each dimension of each student is obtained, wherein the deviation value of each dimension is determined according to the deviation parameter of each dimension, and the deviation parameter is determined according to the accumulated weight value of each dimension and the degree of deviation of each dimension, and the accumulated weight value is determined according to the time corresponding to each stage and the degree of deviation of the same dimension at different stages; The user portrait acquisition management module is used to perform adaptive PCA dimensionality reduction on each student based on the dimension vector and deviation value of each dimension of each student and obtain the user portrait, which is stored in the information data comprehensive management platform for collaborative education between home, school and community.

Images