Knowledge point mastering degree identification method, device, terminal equipment and storage medium

Abstract

The present application is applicable to the field of computer-assisted educational technology, and provides a method, device, terminal device and storage medium for identifying the degree of mastery of knowledge points. The method includes: obtaining exercise data of multiple users respectively; establishing potential scoring matrices of multiple users and multiple knowledge points respectively according to the exercise data; establishing user feature matrices and knowledge point feature matrices that obey the target distribution respectively; updating the potential scoring matrix through the user feature matrix and the knowledge point feature matrix to obtain the target scoring matrix; identifying the degree of mastery of the target knowledge point by the target user according to the target scoring matrix and the data of the target user practicing the exercises of the target knowledge point. The above method is used to establish a user feature matrix and a knowledge point feature matrix that obey the normal distribution to update the potential scoring matrix, so that the theory of the updated target scoring matrix can be rigorous, and the accuracy of identifying the degree of mastery of the user of the target knowledge point can be higher.

Description

Technical Field

[0001] The present application belongs to the field of computer-assisted education technology, and in particular relates to a method, apparatus, terminal device and storage medium for identifying the degree of mastery of a knowledge point. Background Art

[0002] In the traditional evaluation process of subject knowledge points, it is usually based on the corresponding knowledge points learned by students, and the test questions of the corresponding knowledge points are sampled to evaluate the students' mastery of the knowledge points. For example, most of the strategies adopted are test scores to evaluate the students' mastery of the corresponding knowledge points. However, due to the limitations of the test questions themselves, it is impossible to achieve a detailed evaluation of the mastery of each knowledge point of each student. This evaluation method is too dependent on the students' test scores and does not take into account the students' actual learning situation (for example, the usual problem-solving situation). So in the prior art, there is a problem that it is impossible to scientifically and accurately evaluate the students' mastery of knowledge points. Summary of the Invention

[0003] The embodiments of the present application provide a method, apparatus, terminal device and storage medium for identifying the degree of mastery of knowledge points, which can solve the problem in the prior art that it is impossible to scientifically and accurately evaluate the degree of mastery of students and knowledge points.

[0004] In a first aspect, an embodiment of the present application provides a method for identifying the degree of mastery of a knowledge point, including:

[0005] Acquiring exercise data of a plurality of users respectively, wherein the exercise data includes data of the plurality of users practicing exercises on a plurality of knowledge points respectively;

[0006] Based on the exercise data, establishing potential scoring matrices for the multiple users and the multiple knowledge points respectively;

[0007] Establishing user feature matrices for the multiple users and knowledge point feature matrices for the multiple knowledge points, respectively, wherein both the user feature matrices and the knowledge point feature matrices obey a target distribution;

[0008] The potential rating matrix is ​​updated by using the user feature matrix and the knowledge point feature matrix to obtain a target rating matrix;

[0009] The target user's mastery of the target knowledge point is identified based on the target scoring matrix and data on the target user's practice of exercises on the target knowledge point.

[0010] In one embodiment, the step of respectively obtaining exercise data of multiple users includes:

[0011] For any exercise in any knowledge point, count the time each user spends practicing the exercise in that knowledge point;

[0012] If the time taken to complete the exercise is less than the preset time, the data of the user practicing the exercise on the knowledge point is deleted;

[0013] If the time duration for doing the exercises is greater than or equal to the preset time duration, the data of the user practicing the exercises on the knowledge point is determined as the exercise data.

[0014] In one embodiment, establishing potential rating matrices for the plurality of users and the plurality of knowledge points based on the exercise data includes:

[0015] For any user, the data of the user's practice of the exercises on the multiple knowledge points are input into a preset knowledge point scoring model to obtain the user's mastery of the multiple knowledge points;

[0016] According to the mastery degree of the multiple knowledge points, potential rating matrices of the user and the multiple knowledge points are respectively constructed.

[0017] In one embodiment, the potential rating matrix includes an indicator function; and constructing the potential rating matrices for the user and the plurality of knowledge points based on the mastery levels of the plurality of knowledge points includes:

[0018] For the mastery degree of any knowledge point, if the mastery degree of the knowledge point is greater than a preset value, the output value of the indicator function is set as a first target value, and the first target value is used to enhance the user's mastery degree of the knowledge point; or

[0019] For the mastery level of any knowledge point, if the mastery level of the knowledge point is less than or equal to a preset value, the output value of the indicator function is set as a second target value, and the second target value is used to weaken the user's mastery level of the knowledge point;

[0020] The potential scoring matrix is ​​constructed according to the first target value or the second target value.

[0021] In one embodiment, the updating of the potential rating matrix using the user feature matrix and the knowledge point feature matrix to obtain a target rating matrix includes:

[0022] Constructing an update function through the user feature matrix, the knowledge point feature matrix and the potential score matrix;

[0023] The potential scoring matrix is ​​updated according to the update function to obtain a target scoring matrix.

[0024] In one embodiment, the update function is a joint distribution function; and constructing the update function using the user feature matrix, the knowledge point feature matrix, and the potential rating matrix includes:

[0025] The user feature matrix, the knowledge point feature matrix and the potential score matrix are subjected to matrix multiplication to obtain the joint distribution function.

[0026] In one embodiment, updating the potential scoring matrix according to the update function to obtain a target scoring matrix includes:

[0027] Calculating the maximum value in the update function according to the maximum likelihood estimation algorithm;

[0028] Based on the maximum value, the potential score matrix is ​​updated by a gradient descent method to obtain the target score matrix.

[0029] In a second aspect, an embodiment of the present application provides a device for identifying the degree of mastery of a knowledge point, comprising:

[0030] An acquisition module, configured to respectively acquire exercise data of a plurality of users, wherein the exercise data includes data of the plurality of users practicing exercises on a plurality of knowledge points;

[0031] A first establishing module is used to establish potential scoring matrices for the plurality of users and the plurality of knowledge points respectively according to the exercise data;

[0032] A second establishing module is configured to respectively establish user feature matrices of the plurality of users and knowledge point feature matrices of the plurality of knowledge points, wherein the user feature matrices and the knowledge point feature matrices respectively obey a target distribution;

[0033] An updating module, configured to update the potential rating matrix based on the user feature matrix and the knowledge point feature matrix to obtain a target rating matrix;

[0034] The identification module is used to identify the target user's mastery of the target knowledge point based on the target scoring matrix and the data of the target user practicing the exercises on the target knowledge point.

[0035] In a third aspect, an embodiment of the present application provides a terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method described in any one of the first aspects above is implemented.

[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of the first aspects above is implemented.

[0037] In a fifth aspect, an embodiment of the present application provides a computer program product, which, when executed on a terminal device, enables the terminal device to execute any of the methods described in the first aspect above.

[0038] Compared with the prior art, the beneficial effect of the embodiment of the present application is that by obtaining the exercise data of multiple users to establish a potential rating matrix, the problem of model overfitting when the potential rating matrix is ​​established based on a small number of answers of a single user in the past can be solved. Afterwards, a user feature matrix and a knowledge point feature matrix that obey the normal distribution are established respectively. In this way, when the potential rating matrix is ​​updated based on the matrix that obeys the normal distribution, the obtained target rating matrix can be made theoretically rigorous and have stronger interpretability, and the scoring of the degree of mastery of the knowledge points based on this theory is more scientific. Based on this, when the terminal device identifies the target knowledge point based on the exercise data, the accuracy of the terminal device in identifying the target user's mastery of the target knowledge point can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0040] Figure 1 This is a flowchart of a method for identifying the mastery level of a knowledge point provided in one embodiment of the present application;

[0041] Figure 2 This is a schematic diagram of an implementation of S101 of a method for identifying the mastery level of a knowledge point provided in an embodiment of the present application;

[0042] Figure 3 This is a schematic diagram of an implementation method of S102 of a method for identifying the mastery level of a knowledge point provided in an embodiment of the present application;

[0043] Figure 4 This is a schematic diagram of an implementation method of S1022 of a method for identifying the mastery level of a knowledge point provided in an embodiment of the present application;

[0044] Figure 5 This is a schematic diagram of an implementation method of S104 of a method for identifying the mastery level of a knowledge point provided in an embodiment of the present application;

[0045] Figure 6 This is a schematic diagram of an implementation method of S1042 of a method for identifying the mastery level of a knowledge point provided in an embodiment of the present application;

[0046] Figure 7 This is a structural block diagram of a knowledge point mastery level identification device provided by an embodiment of the present application;

[0047] Figure 8 This is a structural block diagram of a terminal device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0048] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0049] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0050] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0051] The knowledge point mastery level identification method provided in the embodiment of the present application can be applied to terminal devices such as mobile phones, tablet computers, laptops, ultra-mobile personal computers (UMPCs), and netbooks. The embodiment of the present application does not impose any restrictions on the specific type of terminal devices.

[0052] See also Figure 1 , Figure 1 The following is a flowchart illustrating a method for identifying the mastery level of a knowledge point provided in an embodiment of the present application. The method includes the following steps:

[0053] S101. Acquire exercise data of a plurality of users respectively, wherein the exercise data includes data of the plurality of users practicing exercises on a plurality of knowledge points respectively.

[0054] In the application, the above-mentioned exercise data is data for users to practice exercises on multiple knowledge points respectively. Among them, the knowledge points can be knowledge points set in advance by the staff in the terminal device. For example, exercises can be divided into exercises in subjects such as mathematics, English, and physics. Each subject can be divided into multiple grades, and each grade contains corresponding knowledge points that need to be learned, which can be consistent with the knowledge points in the corresponding textbooks. It is understandable that the exercises can be pre-cached in the terminal device by the staff based on the corresponding knowledge points. A large number of questions that meet the knowledge points are cached in advance in the terminal device. Among them, the above-mentioned exercise data includes but is not limited to data such as the time the user practices the exercise, whether the answer is accurate, and the difficulty of the exercise, and this is not limited to this.

[0055] In the application, the above-mentioned acquisition of the user's exercise data can be used by the user to practice exercises using the teaching device, and the teaching device can directly upload the acquired exercise data to the real-time data warehouse of the terminal device for caching the exercise data. Exemplarily, the above-mentioned teaching device can be an educational tablet terminal, and the above-mentioned terminal device can be a teaching system or teaching terminal that has a data connection relationship with the educational tablet terminal, or the teaching system includes an educational tablet terminal, which is not limited to this. At this time, the above-mentioned exercise data is generated on the teaching device and then sent to the terminal device. Specifically, the user can log in to the account on the educational tablet terminal in advance, and then, when the user uses the educational tablet terminal to practice exercises, the educational tablet terminal can record the user's exercise data. Finally, the educational tablet terminal can upload the exercise data to the real-time data warehouse of the teaching terminal (terminal device) for caching. Among them, the uploaded exercise data also includes user data, that is, the account number that the user logs in to the educational tablet terminal, which is used to uniquely identify the user.

[0056] It is understood that the above-mentioned users include not only school students, but also other users who need to take exams. For example, if a user who has already started working needs to take an exam for a corresponding certificate, he or she can also download the corresponding teaching app on the above-mentioned educational tablet terminal to practice exercises, without limitation.

[0057] It should be noted that many current methods for predicting knowledge mastery primarily utilize existing exercise data to build learning models, and then predict the user's mastery of a particular knowledge point based on the resulting learning model. However, for a single user, the amount of existing exercise data is often severely insufficient. A learning model trained on this exercise data typically cannot accurately output the user's mastery of a particular knowledge point. Therefore, it is possible to obtain exercise data from multiple users and use it to update a matrix model, so that the resulting matrix model can accurately output the user's mastery of a particular knowledge point.

[0058] S102: establishing potential rating matrices for the multiple users and the multiple knowledge points based on the exercise data.

[0059] In the application, for multiple knowledge points, a potential rating matrix for each user relative to each knowledge point can be constructed. Among them, any potential rating matrix is ​​used to represent the user's mastery of the corresponding knowledge point. For example, for a user's exercise data relative to each knowledge point, the exercise data of each knowledge point can be first input into the existing learning model. Afterwards, the learning model extracts the exercise features of the exercise data for model processing, and then outputs the user's mastery of each knowledge point. After obtaining the mastery level, the terminal device can establish a potential rating matrix for the user and each knowledge point based on the mastery level. It can be understood that the potential rating matrix contains unknown matrix parameters, which can be updated and determined during the subsequent corresponding matrix processing to obtain the final target rating matrix.

[0060] In other applications, the above-mentioned mastery level can also be determined by the accuracy of the exercise data for each knowledge point. For example, it has been described in S101 that the above-mentioned exercise data also includes whether the user's answer to the exercise is accurate. Based on this, one knowledge point corresponds to multiple exercises. When the user practices multiple exercises, the accuracy of the user's practice of multiple exercises can be obtained. For example, the terminal device can count the number of exercises practiced by the user and the number of exercises answered accurately, and then calculate the accuracy of the user's practice of the exercises for this knowledge point. Finally, the terminal device can use the accuracy rate as the user's mastery level of the knowledge point, reducing the process of model training required by the terminal device and the memory of the terminal device required by the model.

[0061] S103 , establishing user feature matrices for the multiple users and knowledge point feature matrices for the multiple knowledge points respectively, wherein both the user feature matrices and the knowledge point feature matrices obey a target distribution.

[0062] In application, the above-mentioned target distribution includes but is not limited to Gaussian distribution (normal distribution), chi-square distribution, etc. In this embodiment, the existing educational statistics show that the students' intelligence level, including learning ability, practical hands-on ability, etc., is normally distributed. Therefore, the normal test score distribution should basically obey the normal distribution. It can be considered that for multiple users, the user's mastery of knowledge points should also conform to the normal distribution. Therefore, in this embodiment, the above-mentioned user feature matrix and knowledge point feature matrix can both obey the normal distribution. Furthermore, when the matrix obeying the normal distribution is subjected to matrix processing to obtain the target scoring matrix, the theory of the target scoring matrix obtained is rigorous and has stronger interpretability, and the score of the knowledge point mastery degree implemented based on this theory is more scientific. Compared with using a neural network model to predict the mastery degree of users and knowledge points, the neural network model generally has low interpretability, not only seriously occupies the computing resources of the terminal device, but also is not suitable for rigorous teaching scenarios.

[0063] Specifically, the terminal device can pre-set U as the user feature matrix of the user, U∈R D*N Similarly, the terminal device can pre-set V as the knowledge point feature matrix of the knowledge point, V∈R D*M Where D is the parameter that needs to be adjusted, M is the number of knowledge points, N is the number of users, and R is the potential rating matrix between users and knowledge points. Since both the user feature matrix and the knowledge point feature matrix obey the normal distribution, the user feature matrix U∈R D*N Subject to μ=0, standard deviation is σ u Similarly, we can set the knowledge point feature matrix V∈R D*M Subject to μ=0, standard deviation is σ v Based on this, the matrix formula of the user feature matrix can be established: Among them, I is the preset indicator function, Indicates the standard deviation as σ u The user feature matrix of the i-th user; and the matrix formula for establishing the knowledge point feature matrix: Where, I is a preset indicator function, where Indicates the standard deviation as σ v The knowledge point feature matrix of the j-th knowledge point. When setting the initial value of μ, it can also be other values ​​(non-zero values), and there is no limitation on this.

[0064] S104: Update the potential rating matrix using the user feature matrix and the knowledge point feature matrix to obtain a target rating matrix.

[0065] In applications, updating the latent rating matrix can be performed by updating the unknown parameters in the latent rating matrix to obtain a target rating matrix. For the existing user matrix, knowledge point feature matrix, and latent rating matrix, a joint distribution function can be established. The joint distribution function is then parameterized using maximum likelihood estimation to optimize and update the unknown parameters of the three matrices, resulting in the target rating matrix.

[0066] It should be added that the above joint distribution function can be obtained by performing matrix multiplication processing on the user feature matrix, the knowledge point feature matrix and the potential rating matrix.

[0067] S105 : Identify the target user's mastery of the target knowledge point based on the target scoring matrix and the data of the target user practicing exercises on the target knowledge point.

[0068] In the application, the target user can be any of the multiple users in S101 or any other user, without limitation. It is understood that for multiple knowledge points, the terminal device can obtain the target user's exercise data relative to the target knowledge point and input the exercise data into the target scoring matrix for calculation to identify the target user's mastery of the target knowledge point.

[0069] It is understood that the data on the target user's practice of the target knowledge point exercises includes, but is not limited to, the total time the target user spent practicing the target knowledge point exercises, the accuracy of the exercises, the overall difficulty of the exercises, and other data. Furthermore, the terminal device can perform unified dimensional processing on the aforementioned data, such as the time spent practicing the exercises, the accuracy of the exercises, and the overall difficulty of the exercises, and then obtain the corresponding numerical values ​​for matrix operations.

[0070] In this embodiment, by obtaining the exercise data of multiple users to establish a potential rating matrix, the problem of model overfitting when the potential rating matrix is ​​established based on a small number of answers of a single user in the past can be solved. Afterwards, a user feature matrix and a knowledge point feature matrix that obey the normal distribution are established respectively. In this way, when the potential rating matrix is ​​updated based on the matrix that obeys the normal distribution, the obtained target rating matrix can be made theoretically rigorous and more interpretable, and the scoring of the degree of mastery of the knowledge points based on this theory is more scientific. Based on this, when the terminal device identifies the target knowledge point based on the exercise data, the accuracy of the terminal device in identifying the target user's mastery of the target knowledge point can be improved.

[0071] Please refer to Figure 2 In a specific embodiment, S101 obtains exercise data of multiple users respectively, which specifically includes the following sub-steps S1011-S1013, which are detailed as follows:

[0072] S1011. For any exercise in any knowledge point, count the time each user spends practicing the exercise in the knowledge point.

[0073] In the application, the above-mentioned question-solving time is the time it takes for the user to practice any exercises on the knowledge points. For example, after the user logs in to the education APP of the education tablet terminal, a plurality of knowledge points will usually be displayed in the interface of the education APP. Afterwards, the knowledge points are selected according to the selection instructions entered by the user on the interface to practice the exercises. However, in the actual process, the user may make an accidental touch and select the wrong knowledge point. At this time, the education tablet terminal can use the time when the exercises are displayed in the interface as the starting time, and the time when the user exits the question-solving interface (reselects the knowledge point or exits the education APP) as the end time. In this way, the education tablet terminal can obtain the time it takes for the user to practice the exercises for each knowledge point.

[0074] It is understandable that when a user accidentally selects an exercise on an incorrect knowledge point for practice, the time the user spends practicing the exercise on that knowledge point is usually relatively short.

[0075] In other applications, the educational APP of the educational tablet terminal also usually has the function of randomly pushing exercises of any knowledge point. Based on this, the educational tablet terminal can also record the exercise time of the user for each exercise. Afterwards, the exercise time of multiple exercises belonging to the same knowledge point is added up, and this is used as the exercise time for the user to practice the exercises of the knowledge point. It should be added that for the exercises of any knowledge point that are randomly pushed, the pushed exercises may be beyond the user's learning scope (knowledge points that have been learned). Based on this, when the user faces the exercises of the knowledge point in this situation, the user may quickly and arbitrarily select or fill in the answers. Therefore, after recording the exercise time of the user for each exercise, the educational tablet terminal can also delete the relevant exercise data whose exercise time is less than the preset exercise time. Afterwards, the relevant exercise data whose exercise time is greater than or equal to the preset exercise time is retained, and there is no limitation on this.

[0076] S1012. If the time for doing the exercises is less than the preset time, the data of the user practicing the exercises on the knowledge point is deleted.

[0077] S1013. If the time for doing the exercises is greater than or equal to the preset time, the data of the user practicing the exercises on the knowledge points is determined as the exercise data.

[0078] In the application, the above preset duration can be set by the user according to actual conditions, which is used to enable the terminal device to determine whether the knowledge point of the exercise currently being done by the user is generated by the user's accidental touch operation.

[0079] It's understandable that when the duration of a question is less than the preset time, it indicates that the user may have been randomly answering the questions, or may have accidentally selected the wrong knowledge point. Based on this, the data on the user's practice of the exercises for that knowledge point can be considered to be of no reference value. Therefore, to make the established potential scoring matrix more reasonable and closer to actual conditions, it is necessary to delete the data on exercises that took less than the preset time, retaining only the data on users who practiced the exercises for the knowledge point for a time greater than or equal to the preset time.

[0080] In other applications, when educational tablets upload exercise data to the terminal device's real-time data warehouse, there may be anomalies or missing data. For example, the uploaded exercise data may lack the user's accuracy rate for practicing exercises related to a particular knowledge point, or the duration of time the user spent practicing exercises related to that knowledge point. In these cases, the terminal device must further filter the acquired exercise data to ensure its usability.

[0081] Please refer to Figure 3 In a specific embodiment, S102 establishes potential rating matrices for the multiple users and the multiple knowledge points based on the exercise data, and further includes the following sub-steps S1021-S1022, which are described in detail below:

[0082] S1021. For any user, the data of the user's practice of the exercises on the multiple knowledge points are input into a preset knowledge point scoring model to obtain the user's mastery of the multiple knowledge points.

[0083] In the application, the above-mentioned knowledge point scoring model can be a model obtained by the user through pre-training based on the existing exercise data, which is used to predict the user's mastery of the knowledge points based on the currently acquired exercise data. Specifically, for any set of training data (exercise data), the user can pre-set the actual mastery level corresponding to each set of training data. Afterwards, the training data can be input into the initial knowledge point scoring model to obtain the predicted mastery level. Then, the loss value is calculated based on the actual mastery level and the predicted mastery level. Finally, the model parameters in the initial knowledge point scoring model are iteratively updated according to the loss value to obtain the final knowledge point scoring model.

[0084] It is understood that when training the aforementioned knowledge point scoring model, features such as exercise accuracy, exercise duration, and exercise difficulty in the exercise data can be utilized for training. Based on this, using the knowledge point scoring model to determine a user's mastery of a knowledge point is more accurate than using the user's accuracy rate for the exercises under that knowledge point to determine the user's mastery of the knowledge point.

[0085] It should be noted that the aforementioned knowledge point scoring model is used to initially output a user's mastery of a knowledge point. The terminal device can then establish an initial latent feature matrix based on this mastery level. Based on this, the latent feature matrix is ​​further updated based on the user feature matrix and the knowledge point feature matrix. This results in a more accurate target scoring matrix, improving the accuracy of identifying the target user's mastery of the target knowledge point.

[0086] S1022. Construct potential rating matrices for the user and the multiple knowledge points respectively according to the mastery levels of the multiple knowledge points.

[0087] In applications, after determining the mastery level of multiple knowledge points, a potential scoring matrix that obeys the normal distribution can also be established, so that the obtained target scoring matrix is ​​theoretically rigorous and has stronger interpretability.

[0088] Specifically, the potential rating matrix R can be set to obey a Gaussian distribution (normal distribution) with a mean of μ and a standard deviation of σ, and its expression can be: Where Iij is an indicator function, which means that if the degree of mastery of the knowledge point Vj by the student Ui meets the expected value, then Iij is 1, otherwise it is 0. 2 ) represents the potential rating matrix with a standard deviation of σ; T represents the transpose step performed on the matrix Ui to obtain the transposed matrix of Ui.

[0089] It should be noted that for multiple users and multiple knowledge points, a potential rating matrix for each user and each knowledge point needs to be established. That is, according to the formula on the right side of the above expression, it can be considered that the above p(RU, V, σ 2 ) is represented as the potential rating matrix of the i-th user and the j-th knowledge point.

[0090] Please refer to Figure 4 In a specific embodiment, the potential rating matrix includes an indicator function; S1022 constructs potential rating matrices for the user and the multiple knowledge points based on the mastery of the multiple knowledge points, and further includes the following sub-steps S10221-S10223, which are described in detail below:

[0091] S10221: For any knowledge point, if the mastery level of the knowledge point is greater than a preset value, the output value of the indicator function is set as a first target value, and the first target value is used to enhance the user's mastery level of the knowledge point. Or,

[0092] S10222. For the mastery level of any knowledge point, if the mastery level of the knowledge point is less than or equal to a preset value, the output value of the indicator function is set to a second target value, and the second target value is used to weaken the user's mastery level of the knowledge point.

[0093] S10223. Construct the potential scoring matrix according to the first target value or the second target value.

[0094] In the application, the above-mentioned preset value, first target value and second target value can all be values ​​set by the user according to actual conditions, and their values ​​can be changed by the user without limitation.

[0095] Specifically, the expression of the potential scoring matrix R in S1022 above can be referred to. The Iij in the expression is the indicator function, which means that if the mastery of the knowledge point Vj by the student Ui meets the expected value, then Iij is 1, otherwise it is 0. At this time, the expected value is the preset value, 1 is the first target value, and 0 is the second target value. A potential scoring matrix is ​​established based on the value output by the indicator function, which can be used to better express the user's mastery of the knowledge point. Furthermore, the value (predicted mastery) finally output by the potential scoring matrix is ​​closer to the actual value (actual mastery). That is, after updating the feature matrix (potential scoring matrix) established according to the indicator function through the user feature matrix and the knowledge point feature matrix, the updated target scoring matrix can be used to estimate the mastery degree of the user and the knowledge point according to the exercise data. The accuracy is higher (that is, closer to the user's actual mastery of the knowledge point).

[0096] Please refer to Figure 5 In a specific embodiment, S104 updates the potential rating matrix using the user feature matrix and the knowledge point feature matrix to obtain a target rating matrix, and further includes the following sub-steps S1041-S1042, which are described in detail below:

[0097] S1041. Construct an update function using the user feature matrix, the knowledge point feature matrix, and the potential rating matrix.

[0098] In applications, the update function may specifically be a joint distribution function, which may be obtained by performing matrix multiplication on the user feature matrix, the knowledge point feature matrix, and the potential rating matrix.

[0099] Specifically, please refer to the detailed description in S103 and S1022 above. Since the above user feature matrix, knowledge point feature matrix and potential score matrix are all matrices that obey the normal distribution, the joint distribution of the above three matrices R (potential score matrix), U (user feature matrix) and V (knowledge point feature matrix) can be derived based on Bayesian theory. That is, the above three matrices are multiplied together, and the expression of their joint distribution function is as follows:

[0100] S1042: Update the potential scoring matrix according to the update function to obtain a target scoring matrix.

[0101] In practice, the joint distribution function expression in S1041 above indicates that it is obtained by multiplying three matrices. Based on this, it can be considered that updating the update function simultaneously updates all three matrices. In other words, updating the update function includes updating the potential score matrix to obtain the target score matrix.

[0102] In applications, when updating the update function, a maximum likelihood estimation algorithm or a Bayesian estimation algorithm may be used for processing, and there is no limitation on this.

[0103] Please refer to Figure 6 In a specific embodiment, S1042 updates the potential score matrix according to the update function to obtain a target score matrix, and further includes the following sub-steps S10421-S10422, which are described in detail below:

[0104] S10421. Calculate the maximum value in the update function according to the maximum likelihood estimation algorithm.

[0105] S10421. Based on the maximum value, update the potential score matrix by a gradient descent method to obtain the target score matrix.

[0106] In the application, the format of the joint distribution function expression in S1042 is consistent with that of the likelihood function expression. Based on this, the maximum likelihood estimation algorithm can be used to solve the parameters of the joint distribution function, and then the joint distribution function is updated (the potential score matrix is ​​updated) according to the solved parameters.

[0107] Specifically, when calculating the maximum value in the update function, the above joint distribution function can be logarithmized to obtain the following expression:

[0108]

[0109]

[0110]

[0111] Wherein, C is a constant, and the meanings of the remaining letters can be specifically referred to the descriptions in S103 and S1022 above, which will not be explained here again. Afterwards, the parameter solution problem of the above formula can be transformed into a method of finding the minimum value of a function in mathematics. Specifically, the above formula can be specifically converted to:

[0112]

[0113] in, At this time, since the logarithmic functions (A function and B function) both carry a "-", the converted L function does not add a "-". Based on this, it can be considered that after finding the minimum value of the L function, the minimum value of the L function, after adding a "-", is the maximum value of the logarithmic function mentioned above. After that, the parameters can be updated based on the maximum value through the gradient descent method to obtain the parameter σ 2 , Based on this, the updated target score matrix can be obtained according to the above parameters and the expressions in S103 and S1022.

[0114] It should be noted that performing logarithmic processing on the update function, then transforming the logarithmic function, and finding the minimum value of the transformed function has the effect of simplifying the process of solving the joint distribution function (calculating the maximum value).

[0115] In application, the above-mentioned gradient descent method includes but is not limited to batch gradient descent, stochastic gradient descent, and mini-batch gradient descent, which are not limited to this. In this embodiment, because the batch gradient descent method uses all samples (all exercise data) to update parameters when updating, the batch gradient descent method can be used to update the potential score matrix to improve the accuracy of the target score matrix in identifying the user and the knowledge point mastery level.

[0116] See also Figure 7 , Figure 7 This is a structural block diagram of a knowledge point mastery level identification device provided in an embodiment of the present application. The modules included in the knowledge point mastery level identification device in this embodiment are used to perform Figures 1 to 6 Each step in the corresponding embodiment. Please refer to Figures 1 to 6 as well as Figures 1 to 6 For the convenience of explanation, only the parts related to this embodiment are shown. Figure 7 The knowledge point mastery level identification device 700 includes: an acquisition module 710, a first establishment module 720, a second establishment module 730, an update module 740 and an identification module 750, wherein:

[0117] The acquisition module 710 is used to respectively acquire exercise data of multiple users, wherein the exercise data includes data of the multiple users practicing exercises on multiple knowledge points.

[0118] The first establishing module 720 is configured to establish potential rating matrices for the plurality of users and the plurality of knowledge points respectively according to the exercise data.

[0119] The second establishing module 730 is used to respectively establish user feature matrices of the multiple users and knowledge point feature matrices of the multiple knowledge points, wherein the user feature matrices and the knowledge point feature matrices respectively obey target distribution.

[0120] The updating module 740 is configured to update the potential rating matrix based on the user feature matrix and the knowledge point feature matrix to obtain a target rating matrix.

[0121] The identification module 750 is configured to identify the target user's mastery of the target knowledge point based on the target scoring matrix and data on the target user's practice of exercises on the target knowledge point.

[0122] In one embodiment, the acquisition module 710 is further configured to:

[0123] For any exercise in any knowledge point, count the time each user spends practicing the exercise in that knowledge point;

[0124] If the time taken to complete the exercise is less than the preset time, the data of the user practicing the exercise on the knowledge point is deleted;

[0125] If the time duration for doing the exercises is greater than or equal to the preset time duration, the data of the user practicing the exercises on the knowledge point is determined as the exercise data.

[0126] In one embodiment, the first establishing module 720 is further configured to:

[0127] For any user, the data of the user's practice of the exercises on the multiple knowledge points are input into a preset knowledge point scoring model to obtain the user's mastery of the multiple knowledge points;

[0128] According to the mastery degree of the multiple knowledge points, potential rating matrices of the user and the multiple knowledge points are respectively constructed.

[0129] In one embodiment, the potential score matrix includes an indicator function; the first building module 720 is further configured to:

[0130] For the mastery degree of any knowledge point, if the mastery degree of the knowledge point is greater than a preset value, the output value of the indicator function is set as a first target value, and the first target value is used to enhance the user's mastery degree of the knowledge point; or

[0131] For the mastery level of any knowledge point, if the mastery level of the knowledge point is less than or equal to a preset value, the output value of the indicator function is set as a second target value, and the second target value is used to weaken the user's mastery level of the knowledge point;

[0132] The potential scoring matrix is ​​constructed according to the first target value or the second target value.

[0133] In one embodiment, the update module 740 is further configured to:

[0134] Constructing an update function through the user feature matrix, the knowledge point feature matrix and the potential score matrix;

[0135] The potential scoring matrix is ​​updated according to the update function to obtain a target scoring matrix.

[0136] In one embodiment, the update function is a joint distribution function; the update module 740 is further configured to:

[0137] The user feature matrix, the knowledge point feature matrix and the potential score matrix are subjected to matrix multiplication to obtain the joint distribution function.

[0138] In one embodiment, the update module 740 is further configured to:

[0139] Calculating the maximum value in the update function according to the maximum likelihood estimation algorithm;

[0140] Based on the maximum value, the potential score matrix is ​​updated by a gradient descent method to obtain the target score matrix.

[0141] When it is understood that Figure 7 In the structural block diagram of the knowledge point mastery degree identification device shown in FIG, each unit / module is used to perform Figures 1 to 6 The steps in the corresponding embodiment, and for Figures 1 to 6 Each step in the corresponding embodiment has been explained in detail in the above embodiment. Figures 1 to 6 as well as Figures 1 to 6 The relevant descriptions in the corresponding embodiments will not be repeated here.

[0142] Figure 8 This is a structural block diagram of a terminal device provided by another embodiment of the present application. Figure 8As shown, the terminal device 800 of this embodiment includes: a processor 810, a memory 820, and a computer program 830 stored in the memory 820 and executable by the processor 810, such as a program for the method for identifying the mastery level of a knowledge point. When the processor 810 executes the computer program 830, the steps in each embodiment of the method for identifying the mastery level of a knowledge point are implemented, such as Figure 1 Alternatively, the processor 810 executes the computer program 830 to implement the above Figure 7 The functions of each module in the corresponding embodiment are, for example, Figure 7 For details on the functions of modules 710 to 750, please refer to Figure 7 Related description in the corresponding embodiment.

[0143] Exemplarily, the computer program 830 may be divided into one or more units, one or more of which are stored in the memory 820 and executed by the processor 810 to complete the present application. The one or more units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 830 in the terminal device 800.

[0144] The terminal device may include, but is not limited to, a processor 810 and a memory 820. Those skilled in the art will appreciate that Figure 8 This is merely an example of the terminal device 800 and does not constitute a limitation on the terminal device 800 . The terminal device may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0145] The processor 810 may be a central processing unit, or other general-purpose processor, a digital signal processor, an application-specific integrated circuit, an off-the-shelf programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0146] The memory 820 may be an internal storage unit of the terminal device 800, such as a hard disk or memory of the terminal device 800. The memory 820 may also be an external storage device of the terminal device 800, such as a plug-in hard disk, smart memory card, flash memory card, etc. equipped on the terminal device 800. Furthermore, the memory 820 may include both an internal storage unit of the terminal device 800 and an external storage device.

[0147] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for identifying the degree of mastery of knowledge points, characterized in that: include: Respectively acquiring exercise data of a plurality of users, wherein the exercise data includes data of exercises of a plurality of knowledge points respectively practiced by the plurality of users; According to the exercise data, potential scoring matrices for the multiple users and the multiple knowledge points are respectively established, and any of the potential scoring matrices is used to represent the user's mastery of the corresponding knowledge point; Establishing user feature matrices of the plurality of users and knowledge point feature matrices of the plurality of knowledge points respectively, wherein both the user feature matrices and the knowledge point feature matrices obey a target distribution; The potential scoring matrix is ​​updated by using the user feature matrix and the knowledge point feature matrix to obtain a target scoring matrix; According to the target scoring matrix and the data of the target user practicing the exercises of the target knowledge point, the mastery degree of the target user on the target knowledge point is identified.

2. The method for identifying the degree of mastery of knowledge points according to claim 1, characterized in that: The step of respectively obtaining exercise data of multiple users includes: For any exercise in any knowledge point, count the time each user spends practicing the exercise in the knowledge point; If the time for doing the exercises is less than the preset time, deleting the data of the user practicing the exercises on the knowledge point; If the time duration for doing the exercises is greater than or equal to the preset time duration, the data of the user practicing the exercises on the knowledge point is determined as the exercise data.

3. The method for identifying the degree of mastery of knowledge points according to claim 1, characterized in that: The step of establishing potential scoring matrices for the plurality of users and the plurality of knowledge points respectively according to the exercise data includes: For any user, the data of the user practicing the exercises on the multiple knowledge points are respectively input into a preset knowledge point scoring model to obtain the user's mastery of the multiple knowledge points; According to the mastery degree of the multiple knowledge points, potential scoring matrices of the user and the multiple knowledge points are respectively constructed.

4. The method for identifying the degree of mastery of a knowledge point according to claim 3, characterized in that: The potential scoring matrix includes an indicator function; The step of constructing potential scoring matrices for the user and the plurality of knowledge points respectively according to the mastery degree of the plurality of knowledge points includes: For the mastery degree of any knowledge point, if the mastery degree of the knowledge point is greater than a preset value, the output value of the indicator function is set as a first target value, and the first target value is used to enhance the user's mastery degree of the knowledge point; or, For the mastery degree of any knowledge point, if the mastery degree of the knowledge point is less than or equal to a preset value, the output value of the indicator function is set as a second target value, and the second target value is used to weaken the mastery degree of the user on the knowledge point; The potential scoring matrix is ​​constructed according to the first target value or the second target value.

5. The method for identifying the degree of mastery of a knowledge point according to claim 1, characterized in that: The updating of the potential scoring matrix by using the user feature matrix and the knowledge point feature matrix to obtain a target scoring matrix includes: Constructing an update function through the user feature matrix, the knowledge point feature matrix and the potential score matrix; The potential scoring matrix is ​​updated according to the update function to obtain a target scoring matrix.

6. The method for identifying the degree of mastery of a knowledge point according to claim 5, characterized in that: The update function is a joint distribution function; The constructing an update function by using the user feature matrix, the knowledge point feature matrix and the potential score matrix comprises: The user feature matrix, the knowledge point feature matrix and the potential score matrix are subjected to matrix multiplication processing to obtain the joint distribution function.

7. The method for identifying the degree of mastery of a knowledge point according to claim 5 or 6, characterized in that: The updating of the potential scoring matrix according to the updating function to obtain a target scoring matrix includes: Calculating the maximum value in the update function according to the maximum likelihood estimation algorithm; Based on the maximum value, the potential score matrix is ​​updated by a gradient descent method to obtain the target score matrix.

8. A knowledge point mastery degree identification device, characterized in that: include: An acquisition module, used to respectively acquire exercise data of a plurality of users, wherein the exercise data includes data of exercises of a plurality of knowledge points practiced by the plurality of users; A first establishing module is used to establish potential scoring matrices for the plurality of users and the plurality of knowledge points respectively according to the exercise data, wherein any of the potential scoring matrices is used to represent the mastery degree of the user on the corresponding knowledge point; A second establishing module is used to establish user feature matrices of the multiple users and knowledge point feature matrices of the multiple knowledge points, respectively, wherein the user feature matrices and the knowledge point feature matrices respectively obey the target distribution; An updating module, used for updating the potential scoring matrix based on the user feature matrix and the knowledge point feature matrix to obtain a target scoring matrix; The identification module is used to identify the mastery degree of the target user on the target knowledge point according to the target scoring matrix and the data of the target user practicing the exercises on the target knowledge point.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images