A learning situation diagnosis method, device, equipment and storage medium

Abstract

The present disclosure relates to a learning situation diagnosis method, device, equipment and storage medium. The learning situation diagnosis method provided by the present disclosure comprises: obtaining test question information of a diagnosis test question and answer data of a target object for the diagnosis test question; generating a knowledge point difficulty level matrix of a target knowledge point corresponding to the diagnosis test question according to the test question information and the answer data; and predicting the mastery of the target knowledge point by the target object based on the knowledge point difficulty level matrix and a knowledge point discrimination matrix through a constructed level response model to obtain a knowledge point ability value, wherein the knowledge point discrimination matrix is determined according to the diagnosis test question. The learning situation diagnosis method provided by the present disclosure can quickly and accurately obtain a diagnosis result.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a learning situation diagnosis method, device, equipment and storage medium. Background Art

[0002] Currently, educational measurement is a common application scenario in the education sector. For example, by statistically analyzing or modeling student performance data and other learning data, students' abilities can be quantified, allowing for accurate and personalized learning diagnosis of each student. Learning diagnosis, in other words, diagnosing a student's learning progress. Based on the diagnostic results, weak points can be identified, facilitating subsequent targeted instruction and practice to consolidate these weaknesses.

[0003] However, existing learning situation diagnosis methods rely on a large amount of learning situation data. They cannot be effectively applied to small or single sample learning situation data, and the accuracy of the diagnosis results cannot be guaranteed. Summary of the Invention

[0004] In order to solve the above technical problems, the present disclosure provides a learning situation diagnosis method, device, equipment and storage medium, which can quickly and accurately obtain diagnosis results.

[0005] According to one aspect of the present disclosure, a learning situation diagnosis method is provided, comprising:

[0006] Acquiring test question information of the diagnostic test question and the target subject's answer data for the diagnostic test question;

[0007] Generating a knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test questions based on the test question information and the answer data;

[0008] Based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, the target subject's mastery of the target knowledge point is predicted through the constructed level response model to obtain the knowledge point ability value, wherein the knowledge point discrimination matrix is ​​determined according to the diagnostic test questions.

[0009] According to another aspect of the present disclosure, there is provided a learning situation diagnosis device, comprising:

[0010] An acquisition module, configured to acquire test question information of the diagnostic test question and the target subject's answer data for the diagnostic test question;

[0011] A generation module, configured to generate a knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test questions based on the test question information and the answer data;

[0012] A prediction module is used to predict the target subject's mastery of the target knowledge point through a constructed grade response model based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, and obtain the knowledge point ability value, wherein the knowledge point discrimination matrix is ​​determined according to the diagnostic test questions.

[0013] According to another aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing a program, wherein the program comprises instructions that, when executed by the processor, cause the processor to execute the above-mentioned learning situation diagnosis method.

[0014] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute a learning situation diagnosis method.

[0015] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the above-mentioned learning situation diagnosis method when executed by a processor.

[0016] The technical solution provided by the embodiments of the present disclosure has the following advantages over the prior art:

[0017] The learning situation diagnosis method provided by the present disclosure includes: obtaining the test information of the diagnostic test questions and the target subject's answer data for the diagnostic test questions; generating a knowledge point difficulty level matrix for the target knowledge points corresponding to the diagnostic test questions based on the test information and the answer data; based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, predicting the target subject's mastery of the target knowledge points through the constructed grade response model to obtain the knowledge point ability value, wherein the knowledge point discrimination matrix is ​​determined based on the diagnostic test questions. The learning situation diagnosis method provided by the present disclosure can quickly and accurately obtain diagnostic results. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0019] In order to more clearly illustrate the embodiments of the present disclosure 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, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0020] Figure 1 A flowchart of a learning situation diagnosis method provided in an embodiment of the present disclosure;

[0021] Figure 2A flowchart of a learning situation diagnosis method provided in an embodiment of the present disclosure;

[0022] Figure 3 A flowchart of a learning situation diagnosis method provided in an embodiment of the present disclosure;

[0023] Figure 4 A flowchart of a learning situation diagnosis method provided in an embodiment of the present disclosure;

[0024] Figure 5 A schematic diagram of the structure of a learning situation diagnosis device provided in an embodiment of the present disclosure;

[0025] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] In order to more clearly understand the above-mentioned purposes, features and advantages of the present disclosure, embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of protection of the present disclosure.

[0027] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0028] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the description below. It should be noted that the concepts of "first", "second", etc. mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0029] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more".

[0030] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0031] Before explaining the present disclosure in detail, it is preferred to explain the terms involved, among which:

[0032] CTT (Classical Test Theory): also known as the true score theory, is the earliest measurement theory to be formalized in mathematics. It generally evaluates students' abilities by the accuracy rate of the overall test paper.

[0033] IRT (Item Response Theory): A question response theory based on the relationship between student ability and answer accuracy. It is a general term for psychological statistical models.

[0034] GRM (The Grade Response Model): The IRT model requires that the input test scores can only be 0 and 1. When the scores have multiple or continuous values, they need to be converted into score levels, and then the test ability value is calculated for each score level, which is the grade response model.

[0035] BKT (Bayesian knowledge tracing): A method that uses Bayesian networks to track and diagnose student knowledge points in a certain order.

[0036] DKT (Deep Knowledge Tracing): A method for tracking and diagnosing student knowledge points based on a multi-layer LSTM recurrent neural network.

[0037] EKT (Exercise-aware Knowledge Tracing): A method for dynamic knowledge point tracking and diagnosis based on bidirectional LSTM and attention mechanism using question text information and knowledge point information.

[0038] Guass-Hermite: Gauss-Hermite, a method for approximate integrals.

[0039] Learning situation diagnosis: including but not limited to knowledge point ability diagnosis, subject ability value diagnosis and comprehensive ability diagnosis.

[0040] Currently, existing learning status diagnosis methods rely on large amounts of test-taking data. Traditional diagnostic models, such as Classical Test Theory (CTT), Item Response Theory (IRT), and the Grade Response Model (GRM), require parameter estimation for latent variables such as item discrimination, item difficulty, and guess parameters. They also require a large amount of test-taking data and repeated iterations to achieve parameter convergence. On the other hand, knowledge tracing models, such as Bayesian Knowledge Tracing (BKT), Deep Knowledge Tracing (DKT), and Exercise-aware Knowledge Tracing (EKT), can discover relationships between knowledge points and diagnose students' mastery of these points. However, these models require not only a temporal order in the answer records (i.e., students answer questions in the order of the knowledge points) but also require a large number of students to repeatedly practice on a closed set of knowledge points. This not only requires a large amount of data, but also places high demands on the data. Therefore, traditional models are difficult to apply in real-world educational scenarios.

[0041] In addition, for small sample learning data, that is, when students' answer records are insufficient, existing diagnostic methods cannot calculate students' ability values. Small sample learning data mainly manifests as follows: (1) the number of students taking the same test questions is small; (2) the knowledge points involved in the test questions are relatively scattered and random; (3) the scenarios in which students take the questions are diverse and complex, and do not have obvious time series characteristics. When the sample data contains any of the above three types of data, the existing diagnostic model cannot converge parameters or train the model, and the diagnostic model cannot be effectively applied in this educational measurement scenario.

[0042] In response to the above technical problems, the embodiments of the present disclosure provide a learning situation diagnosis method that is applied to educational measurement scenarios and is not affected by the amount of sample data. For small sample or even single sample learning situation data, it can make full use of the existing small amount of answer records and correction results and other answer data, while considering test information such as test difficulty, student answer continuous scores, and knowledge point examination frequency, and conduct a comprehensive diagnosis of students' ability values ​​from a multi-dimensional perspective, including single knowledge point ability value, single subject comprehensive ability value, and multi-disciplinary comprehensive ability value. This will be explained in detail through one or more of the following embodiments.

[0043] Specifically, the learning situation diagnosis method can be executed by a terminal or a server. Specifically, the terminal or the server can evaluate the student's ability value through a grade response model and a scoring model. The execution subject of the training method of the grade response model and the scoring model and the execution subject of the learning situation diagnosis method can be the same or different. Among them, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited to this. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0044] For example, in one application scenario, a server trains a graded response model and a scoring model. A terminal obtains the trained graded response model and scoring model from the server and then uses them to evaluate a student's ability. In another application scenario, a terminal trains a graded response model and a scoring model. Furthermore, the terminal uses the trained graded response model and scoring model to evaluate a student's ability.

[0045] It is understandable that the training methods for the grade response model and the scoring model, and the learning situation diagnosis method provided by the embodiments of the present disclosure are not limited to the several possible scenarios described above. The following is a detailed explanation using the example of a server training the grade response model and the scoring model, and a terminal executing the learning situation diagnosis method. The training process of the grade response model and the scoring model refers to the training process of the neural network model, and will not be repeated here.

[0046] Figure 1 The flowchart of the learning situation diagnosis method provided by the embodiment of the present disclosure specifically includes the following steps: Figure 1 The following steps S110 to S130 are shown:

[0047] S110: Acquire test question information of the diagnostic test question and the target subject's answer data for the diagnostic test question.

[0048] It is understandable that the test information of the diagnostic test questions and the target object's answer data on the diagnostic test questions are obtained. The diagnostic test questions are at least one test question used to diagnose the target object's ability value, that is, the diagnostic test questions can be regarded as a collection of at least one test question used for learning situation diagnosis. The knowledge points corresponding to at least one test question are not exactly the same. The answer data at least includes the test question score, and the answer data can be a small sample learning situation data. The test information of the diagnostic test questions at least includes the test question knowledge point label and the test question difficulty value label. The test question knowledge point label refers to the label of the knowledge point contained in the test question, that is, the knowledge point and difficulty value corresponding to each test question can be known.

[0049] S120 , generating a knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test questions based on the test question information and the answer data.

[0050] The test question information includes a first test question difficulty, which is a test question difficulty value label.

[0051] It can be understood that on the basis of the above S110, a knowledge point difficulty level matrix of the target knowledge point corresponding to the diagnostic test question is generated according to the test question information of the diagnostic test question and the test question score of the diagnostic test question in the answer data. If the diagnostic test question includes multiple test questions and the knowledge points corresponding to each test question are not exactly the same, then for each knowledge point, a knowledge point difficulty level matrix of the knowledge point is generated according to the answer data and test question information of all diagnostic test questions under the knowledge point. That is to say, each knowledge point will have a corresponding knowledge point difficulty level matrix.

[0052] It is understandable that the input of the traditional GRM model is the test score, which usually requires estimating the two latent variables of test discrimination and test difficulty based on a large amount of answer data based on the Expectation-Maximization algorithm (EM) or the Markov Chain Monte Carlo method (MCMC), and finally outputting the student's test ability value by inputting the student's score level. However, in the case of single sample or small sample learning data, the EM or MCMC algorithm cannot estimate the above two parameters, resulting in the inability of the traditional GRM model to calculate the test ability value. The present disclosure solves the above technical problems by calculating the knowledge point difficulty level matrix of small sample learning data.

[0053] Optionally, the generation of the knowledge point difficulty level matrix in S120 may be implemented by the following steps:

[0054] The first test question difficulty is mapped to a second test question difficulty according to a first mapping rule.

[0055] The diagnostic test question scores in the answer data are mapped into score levels based on a first preset threshold.

[0056] A knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test questions is generated according to the difficulty level of the second test question and the score level.

[0057] It is understandable that, for the difficulty of the test questions, the difficulty of the first test question can be regarded as the original test question difficulty of the test question. The difficulty of the first test question can be a text or a numerical value indicating the difficulty, for example, "harder" or "easy" in text form and "1" or "4" in numerical form. When the difficulty of the test question is expressed numerically, the difficulty of the first test question is an increasing sequence of positive integers, for example, the difficulty of the first test question is an increasing sequence of positive integers from 1 to 5. According to the first mapping rule, the difficulty of the first test question is mapped to the difficulty of the second test question. The difficulty of the second test question can be understood as the difficulty of the new test question. The first mapping rule is constructed according to the theoretical distribution of the latent variable of the test question difficulty of the GRM model, that is, the difficulty of the first test question is mapped to a distribution containing positive and negative numbers with a mean of 0 to form a new test question difficulty (the difficulty of the second test question). For specific mapping rules, please refer to the following test question difficulty mapping table 1.

[0058] Table 1:

[0059] Original test difficulty d 1 (easy) 2 (easier) 3 (mid-range) 4 (more difficult) 5 (difficult) <![CDATA[New test difficulty d n > -3 -2 0 2 3

[0060] In the table, the difficulty of the first question is represented by d, and the difficulty of the second question is represented by d n Indicates that the first row of Table 1 contains the original test difficulty d in text form, and the values ​​in the table indicate the difficulty value of the test question.

[0061] It can be understood that for the score level, the first preset threshold is the set maximum score level, and the first preset threshold is recorded as K, where K>=1, indicating that the GRM model is divided into K+1 score levels in total. For example, K=4, a total of 5 score levels are divided. According to prior knowledge, the diagnostic test score (represented by s) in the answer data is mapped to the score level (represented by k). The diagnostic test score can be understood as the student's original answer score. In order to facilitate mapping, the diagnostic test score can be normalized to the range of 0 to 1, and the interval corresponding to each score level is divided within the range of 0 to 1. Then, the diagnostic test score is mapped to the corresponding score level. For specific mapping rules, please refer to the following score level mapping table 2.

[0062] Table 2:

[0063] Student's original score 0-0.2 0.2-0.4 0.4-0.6 0.6-0.8 0.8-1.0 Score level k 0 1 2 3 4

[0064] In the table, the first preset threshold, K, is 4. The GRM model uses five scoring levels, ranging from 0 to 4. The test scores are also divided into five intervals based on the scoring levels, with each interval corresponding to a scoring level. Since the GRM model calculates the probability of a score, and a score of 0 indicates a complete error, questions with a score of 0 are directly eliminated. The scoring levels, k∈[1, 2, 3, 4], are used.

[0065] It is understandable that based on the difficulty and score level of the second test question obtained after mapping, a knowledge point difficulty level matrix of the target knowledge points corresponding to the diagnostic test question is generated. The specific implementation steps are as follows:

[0066] The score level is mapped to a third test question difficulty level according to a second mapping rule.

[0067] A test question difficulty level is obtained according to the difficulty level of the second test question and the difficulty level of the third test question.

[0068] According to the difficulty level of the test questions, a knowledge point difficulty level matrix of the target knowledge points corresponding to the diagnostic test questions is constructed.

[0069] It is understandable that according to the second mapping rule, the score level is mapped to the difficulty of the third test question. In the theoretical distribution of the difficulty of the original GRM model test question, the higher the score level, the greater the theoretical value of the corresponding test question difficulty. The second mapping rule is constructed based on the theoretical distribution of the test question difficulty. Specifically, a mapping function between the score level and the test question difficulty can be designed according to the second mapping rule. The mapping function is shown in formula (1). The score level can be mapped to the difficulty of the third test question. The difficulty of the third test question can be understood as the basic difficulty. After obtaining the difficulty of the third test question through formula (1), the difficulty of the second test question and the difficulty of the third test question are added to obtain the test question difficulty level. The calculation method is shown in the following formula (2).

[0070]

[0071] Where, d b is the difficulty of the third test question; m represents the difficulty threshold of the test question, which is used to control the upper and lower bounds of the difficulty of the GRM model. The default value is m = 4; n represents the difficulty slope of the test question, which is used to control the speed of change of the difficulty of the GRM model. The default value is n = 5. The larger n is, the slower the change of difficulty; k represents the current score level. If the maximum score level K = 4, then k∈[1, 2, 3, 4]; q is the median of the scores of all test questions, where the difference between k and q may be negative.

[0072] d q =d b +d n Formula (2)

[0073] Where, d q is the difficulty level of the test question, dn The difficulty of the second question.

[0074] It is understandable that the number of values ​​of the test question difficulty level and the score level calculated according to the above formula (1) and formula (2) is consistent, that is, both are K elements, and the test question difficulty level is divided into K levels, which are recorded as the first test question difficulty level to the Kth test question difficulty level. After obtaining the test question difficulty level, if the diagnostic test question includes multiple test questions, and the multiple test questions correspond to multiple knowledge points, then the multiple test questions are grouped based on each knowledge point, that is, the test questions including the same knowledge point are grouped into a group, and a knowledge point difficulty level matrix of the knowledge point is constructed. Each knowledge point corresponds to a knowledge point difficulty level matrix, and the knowledge point difficulty level matrix is ​​shown in formula (3). Based on the above example, if the number of test questions under the target knowledge point after grouping is t, then the size of the knowledge point difficulty level matrix of the target knowledge point is t*K, that is, t rows and K columns, K is the maximum score level, each row corresponds to a test question, and each column represents a test question difficulty level d q For example, the first row is [0, 0, 1, 0], and the first row corresponds to the first test question, indicating the test difficulty level d of the first test question calculated by formula (2). q =3, in the knowledge point difficulty level matrix, the difficulty level of the third test question represented by the third column is recorded as 1, and the difficulty levels of different test questions represented by the remaining columns are recorded as 0.

[0075]

[0076] Where β is the knowledge point difficulty level matrix composed of the difficulty levels of the test questions, d q11 Indicates the value of the first question at the first difficulty level, d q1K Indicates the value of the first question at the difficulty level of the Kth question.

[0077] S130 , based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, predict the target subject's mastery of the target knowledge point through the constructed level response model to obtain the knowledge point ability value.

[0078] The knowledge point discrimination matrix is ​​determined based on the diagnostic test questions.

[0079] It can be understood that, based on the above S120, the knowledge point difficulty level matrix and the knowledge point discrimination matrix are input into the grade response model. The grade response model can be understood as the grade response model for small sample learning data (The Grade Response Model for Small Sample Answer Records, GRM-SSAR). The grade response model will predict the target object's mastery of the target knowledge point and output the knowledge point ability value for the target knowledge point. The grade response model can predict the ability value of one knowledge point at a time. Every time a knowledge point difficulty level matrix is ​​calculated, it can be input into the grade response model to calculate the ability value of the knowledge point. For example, the diagnostic test consists of 10 test questions. The 10 test questions are grouped according to the test question knowledge point labels included in the test question information. If they are divided into 2 groups, it means that the 10 test questions contain 2 different knowledge points. According to the test question difficulty levels of all test questions in the knowledge point 1 group, the knowledge point difficulty level matrix 1 of knowledge point 1 is formed, and then the knowledge point discrimination matrix and the knowledge point difficulty level matrix 1 are input into the grade response model to calculate the ability value of knowledge point 1. The calculated ability value reflects the quantified target object's mastery of knowledge point 1.

[0080] As can be understood, if the target subject takes t questions (t>=1) for the target knowledge point, the discrimination between the questions can be set to be the same. In this case, the knowledge point discrimination matrix α = [[1,1,...,1]], with a size of 1*t. This means that the knowledge point discrimination matrix is ​​determined based on the number of diagnostic questions. For the same batch of diagnostic questions, the knowledge point discrimination matrix can be the same.

[0081] Optionally, obtaining the knowledge point capability value in S130 may be implemented by the following steps:

[0082] Obtain the simulation capability value and weight value corresponding to the target object.

[0083] According to the simulated ability value, the knowledge point difficulty level matrix and the knowledge point discrimination matrix, the target subject's mastery of the target knowledge point is predicted through the constructed level response model, and the probability of the target knowledge point being associated with the score level is obtained.

[0084] The knowledge point ability value is obtained according to the probability of the target knowledge point being associated with the score level, the simulated ability value and the weight value.

[0085] Understandably, the simulated ability value and weight value corresponding to the target object are obtained. Generally, the overall ability value of the student obeys the normal distribution. The number of samples is specified, for example, size = 30, and the simulated ability value of the student can be obtained by Gauss-Hermite integration (using θi denoted) and the corresponding weights (denoted by g(θ i ), that is, the simulated ability value and weight value can be randomly determined, and the simulated ability value can be normalized to the range of [-3, 3]. Then, based on the simulated ability value, the knowledge point difficulty level matrix and the knowledge point discrimination matrix determined above, the target subject's mastery of the target knowledge point is predicted through the constructed grade response model, and the probability of the target knowledge point being associated with the score level is obtained. The probability of the target knowledge point being associated with the score level can also be called the score level probability. The calculation formulas for the score level probability are shown in Formulas (4) and (5).

[0086]

[0087] Where, is the probability that the score level is greater than or equal to k, D represents a fixed constant, usually D = 1.702, α represents the knowledge point discrimination matrix, β represents the knowledge point difficulty level matrix, and θ represents the simulation ability value.

[0088]

[0089] Where p k It represents the probability of the target knowledge point being related to the score level, which can also be called the probability of the score level being k. It represents the probability that the score level is greater than or equal to k+1. The probability that the score level is greater than or equal to 1 is greater than the probability that the score level is greater than or equal to 2. Therefore, the probability of the score level being 1 is obtained by subtracting the probability that the score level is greater than or equal to 2 from the probability that the score level is greater than or equal to 1.

[0090] It can be understood that after obtaining the probability of the target knowledge point being related to the score level, the joint probability of the probability of the score level being k is calculated, and the logarithm is taken to obtain the score level log likelihood function, which is specifically calculated by the following formula (6).

[0091]

[0092] Where L is the log-likelihood function of the score level, t is the number of test questions, K is the maximum score level, is the joint probability of the probability of scoring level k, r represents the score list after one-hot encoding of the score level of the target object, and the size of r is 1*K+1. For example, object A has done 1 test question, the score level k=2, and the maximum score level K=4, then r=[0, 0, 1, 0, 0], indicating that the third column value corresponding to k=2 is 1, and the values ​​of the other columns are 0. Object A has done 2 questions, the size of r is 2*5, the score level k=3, r=[0, 0, 0, 1, 0], indicating that only the fourth column value corresponding to k=3 is 1. For example, object B has done 1 test question, and the score level k=4, then r=[0, 0, 0, 0, 1], indicating that the fifth column value corresponding to k=4 is 1.

[0093] It can be understood that according to the log-likelihood function of the score level, the simulated ability value and the weight value, the expected value of the target object's ability value for the target knowledge point is obtained through the Bayesian expected a posteriori estimation (EAPE), that is, formula (7). Then, the idea of ​​Gauss-Hermite integration of formula (8) is used to approximate the solution of formula (7), and the estimated value of the target object's ability value for doing t questions under the target knowledge point can be obtained, as shown in formula (9). If the results of formula (7) and formula (8) are similar, that is, the expected value of the ability value and the estimated value of the ability value are close, the knowledge point ability value is output. The relevant formula is as follows:

[0094]

[0095]

[0096]

[0097] Where, E(θ i ) is the expected value of ability, is the estimated value of the ability value, θ is the simulated ability value, L is the log-likelihood function of the score level, g(θ i ) weight value, size represents the number of samples specified when simulating the normal distribution of capability values, and the default size is 30.

[0098] For example, see Figure 2 , Figure 2 This is a flowchart of the learning situation diagnosis method provided by the embodiment of the present disclosure. Figure 2It includes two stages, namely generating a knowledge point difficulty level matrix and generating a knowledge point ability value. The process of generating a knowledge point difficulty level matrix includes: mapping the difficulty of the first test question to the difficulty of the second test question, mapping the test question score to the score level, and generating a knowledge point difficulty level matrix according to the difficulty level of the second test question, the score level and the test question knowledge point; the process of generating a knowledge point ability value includes: generating a score level probability through the GRM model according to the knowledge point discrimination, the knowledge point difficulty level matrix and the simulated ability value, and generating a knowledge point ability value according to the score level probability and the weight value.

[0099] The learning situation diagnosis method provided by the embodiment of the present disclosure proposes a grade response model for small sample learning situation data. After mapping the test score into a score grade and mapping the difficulty of the first test question into the difficulty available in the GRM model, a mapping function between the score grade and the test question difficulty is designed based on the correlation between the score grade and the test question difficulty, making full use of the existing test question difficulty information of the test question, generating a knowledge point difficulty grade matrix, and providing the possibility for solving the GRM model in the small sample learning situation diagnosis scenario. Subsequently, through the knowledge point discrimination, the knowledge point difficulty grade matrix and the simulated ability value and weight value, based on the original GRM model, the knowledge point ability value of the target object for the test question knowledge point is output, and the obtained knowledge point ability value is not only interpretable, but also can truly reflect the student's knowledge point mastery. Secondly, the test question knowledge point label information is fully utilized to convert the test question ability value into the knowledge point ability value, breaking through the limitation that the GRM model can only calculate the test question ability value. In addition, by constructing a knowledge point discrimination matrix with the same discrimination for all test questions in the same batch, the process of the GRM model using a large number of samples to estimate latent variables is avoided, which makes it possible to solve the GRM model in the scenario of small sample learning situation diagnosis and further speeds up the calculation speed.

[0100] Based on the above embodiments, Figure 3 This is a flowchart of the learning situation diagnosis method provided in the embodiment of the present disclosure. After obtaining the knowledge point ability value, the single subject ability value or the comprehensive ability value of multiple subjects can also be calculated based on at least one knowledge point ability value.

[0101] At present, the existing learning situation diagnosis methods can only diagnose students' mastery of a group of questions or a certain knowledge point, and cannot comprehensively evaluate students' mastery of a certain subject or multiple subjects. In other words, they cannot effectively evaluate students' comprehensive subject abilities, nor can they give an ability value that reflects students' overall subject abilities.

[0102] Calculating a single subject ability value based on at least one knowledge point ability value specifically includes: Figure 3 The following steps S310 to S320 are shown:

[0103] S310: Determine the knowledge point examination frequency corresponding to the subject to be examined where the target knowledge point is located.

[0104] It is understandable that the subject to be examined in which the target knowledge point is located is determined, and the examination frequency of the knowledge points in the subject to be examined in the past three years is counted to obtain the examination frequency of each knowledge point.

[0105] S320: Determine the subject ability value of the target object for the subject to be examined through the constructed scoring model based on the knowledge point ability value and the knowledge point examination frequency.

[0106] Optionally, the above-mentioned S310 of determining the subject ability value can be specifically implemented by the following steps:

[0107] The candidate knowledge point examination frequency is determined among the knowledge point examination frequencies.

[0108] The normalized knowledge point ability values ​​are smoothed and linearly scaled.

[0109] According to the examination frequency of the candidate knowledge points and the processed knowledge point ability values, the subject ability value of the target object for the subject to be examined is determined.

[0110] It can be understood that, based on the above S320, after determining the inspection frequency of each knowledge point in the subject to be inspected, important knowledge points are selected to participate in the calculation of the subject ability value to be inspected, and normalized, the inspection frequency of the candidate knowledge points after screening and normalization is obtained, and the inspection frequency of the candidate knowledge points is recorded as f i For example, to filter the top 20 knowledge points by inspection frequency, the number of candidate knowledge points is determined to be 20, and the inspection frequencies of the 20 candidate knowledge points are normalized to values ​​in the range of [0,1], and the inspection frequencies of all candidate knowledge points are f i After addition, it equals 1.

[0111] It is understandable that the range of the knowledge point ability values ​​obtained above is [-3, 3]. The subject ability values ​​obtained by direct weighted averaging may be negative. Therefore, before calculating the subject ability values, that is, before scoring, the knowledge point ability values ​​are first normalized and scaled to the range of [0, 1]. For details, see formula (10) and formula (11).

[0112]

[0113]

[0114] Where, is the knowledge point ability value, c is the knowledge point ability value The minimum value in , d is the knowledge point ability value The maximum value in; Formula (10) is The extreme values ​​of are smoothed to weaken the influence of outliers, and finally 0-1 linear scaling is performed using formula (11).

[0115] It is understandable that the frequency of candidate knowledge point examinations is used to calculate the expectation of the ability value of the processed knowledge point, and the subject ability value of the target object for the subject to be examined is obtained. That is, based on formula (12), the ability value score of the target object for a single subject is calculated according to the frequency of candidate knowledge point examinations and the ability value of the processed knowledge point.

[0116]

[0117] Where, is the ability value of the knowledge point after processing, f i is the frequency of candidate knowledge point inspection, and Z is the number of candidate knowledge points. That is, the subject ability value is calculated through at least one knowledge point related to the subject to be inspected and the ability value of the knowledge point.

[0118] Optionally, after determining the subject ability value of the target subject for the subject to be examined, the student's comprehensive ability value for multiple subjects including the subject to be examined may be determined, which is specifically achieved through the following steps:

[0119] The comprehensive ability value of the target object for at least one subject to be examined is determined based on the subject ability value and the obtained subject weight value.

[0120] It is understandable that at least one subject to be examined is determined, and the subject capability value of each subject to be examined can be determined specifically according to the above method. The specific determination steps are not repeated here. For example, if there are 5 subjects to be examined, and A = 5, the normalized subject weight value is used to calculate the expectation of the subject capability value of each subject to be examined. The expected sum is not divided by A. The normalized subject weight value range is [0, 1], and the subject capability value range is [-3, 3]. That is, the subject weight value of the subject to be examined and the subject capability value of each subject to be examined are input, and the comprehensive capability value of the target object for multiple subjects is output, as shown in formula (13).

[0121]

[0122] Where θ a is the comprehensive ability value of multiple disciplines, w i is the subject weight value, It is the subject ability value.

[0123] For example, see Figure 4 , Figure 4 This is a flowchart of the learning situation diagnosis method provided by the embodiment of the present disclosure. Figure 4The process includes two stages: generating a single-discipline capability value and generating a comprehensive capability value for multiple disciplines. The step of generating a single-discipline capability value includes: using a scoring model to generate a single-discipline capability value based on the normalized capability value of at least one knowledge point and the knowledge point examination frequency. The step of generating a comprehensive capability value for multiple disciplines includes: generating a comprehensive capability value for multiple disciplines based on the capability value of at least one single discipline and the subject weight values.

[0124] The learning situation diagnosis method provided by the disclosed embodiments can scientifically provide the target subject's single-subject ability value by introducing the frequency of knowledge point examinations in a single subject and calculating the mathematical expectation of the ability value of at least one knowledge point. It can also calculate the subject ability values ​​of multiple students for the same subject and compare the subject ability values ​​of multiple students horizontally to fully understand the learning status of multiple students in the subject. Secondly, by introducing subject weight values ​​and calculating the mathematical expectation of at least one single subject ability value, it can scientifically provide the target subject's comprehensive ability value score for multiple subjects. It can also further perform horizontal comparisons of the comprehensive ability values ​​of multiple subjects among multiple students.

[0125] Figure 5 The structure diagram of the learning situation diagnosis device provided by the embodiment of the present disclosure. The learning situation diagnosis device provided by the embodiment of the present disclosure can execute the processing flow provided by the learning situation diagnosis method embodiment, such as Figure 5 As shown, the learning situation diagnosis device 500 includes an acquisition module 510, a generation module 520 and a prediction module 530, wherein:

[0126] An acquisition module 510 is configured to acquire test information of a diagnostic test question and the target subject's answer data for the diagnostic test question;

[0127] A generating module 520 is configured to generate a knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test question based on the test question information and the answer data;

[0128] The prediction module 530 is used to predict the target subject's mastery of the target knowledge point through the constructed grade response model based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, and obtain the knowledge point ability value, wherein the knowledge point discrimination matrix is ​​determined according to the diagnostic test questions.

[0129] Optionally, the test question information includes the difficulty of the first test question.

[0130] Optionally, the generating module 520 is configured to:

[0131] Mapping the first test question difficulty to a second test question difficulty according to a first mapping rule;

[0132] Mapping the diagnostic test question scores in the answer data into score levels based on a first preset threshold;

[0133] A knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test questions is generated according to the difficulty level of the second test question and the score level.

[0134] Optionally, the generating module 520 is configured to:

[0135] Mapping the score level to a third test question difficulty according to a second mapping rule;

[0136] Obtaining a test question difficulty level according to the difficulty level of the second test question and the difficulty level of the third test question;

[0137] According to the difficulty level of the test questions, a knowledge point difficulty level matrix of the target knowledge points corresponding to the diagnostic test questions is constructed.

[0138] Optionally, the prediction module 530 is configured to:

[0139] Obtaining the simulation capability value and weight value corresponding to the target object;

[0140] According to the simulated ability value, the knowledge point difficulty level matrix and the knowledge point discrimination matrix, the target subject's mastery of the target knowledge point is predicted by a constructed level response model, and the probability of the target knowledge point being associated with the score level is obtained;

[0141] The knowledge point ability value is obtained according to the probability of the target knowledge point being associated with the score level, the simulated ability value and the weight value.

[0142] Optionally, the apparatus 500 is further configured to:

[0143] Determine the knowledge point inspection frequency corresponding to the subject to be inspected where the target knowledge point is located;

[0144] According to the knowledge point ability value and the knowledge point examination frequency, the subject ability value of the target object for the subject to be examined is determined through the constructed scoring model.

[0145] Optionally, the apparatus 500 is further configured to:

[0146] Determining a candidate knowledge point inspection frequency from the knowledge point inspection frequencies;

[0147] Performing smoothing and linear scaling on the normalized knowledge point ability values;

[0148] According to the examination frequency of the candidate knowledge points and the processed knowledge point ability values, the subject ability value of the target object for the subject to be examined is determined.

[0149] Optionally, the apparatus 500 is further configured to:

[0150] The comprehensive ability value of the target object for at least one subject to be examined is determined based on the subject ability value and the obtained subject weight value.

[0151] The device provided in this embodiment has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0152] The exemplary embodiments of the present disclosure further provide an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program being configured to cause the electronic device to perform a method according to an exemplary embodiment of the present disclosure when executed by the at least one processor.

[0153] Exemplary embodiments of the present disclosure further provide a computer program product, including a computer program, wherein when the computer program is executed by a processor of a computer, it is used to cause the computer to perform the method according to the embodiment of the present disclosure.

[0154] refer to Figure 6 , a block diagram of an electronic device 600 that can serve as a server or client of the present disclosure will now be described, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0155] like Figure 6 As shown, the electronic device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for the operation of the electronic device 600 can also be stored in the RAM 603. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0156] Multiple components within electronic device 600 are connected to I / O interface 605, including an input unit 606, an output unit 607, a storage unit 608, and a communication unit 609. Input unit 606 can be any type of device capable of inputting information into electronic device 600. Input unit 606 can receive input numeric or character information and generate key input signals related to user settings and / or function control of the electronic device. Output unit 607 can be any type of device capable of presenting information and may include, but is not limited to, a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. Storage unit 608 may include, but is not limited to, a magnetic disk or an optical disk. Communication unit 609 allows electronic device 600 to exchange information / data with other devices via computer networks such as the Internet and / or various telecommunication networks and may include, but is not limited to, a modem, a network card, an infrared communication device, a wireless communication transceiver and / or a chipset, such as a Bluetooth™ device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0157] The computing unit 601 can be various general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above. For example, in some embodiments, the learning situation diagnosis method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 600 via the ROM 602 and / or the communication unit 609. In some embodiments, the computing unit 601 can be configured to execute the learning situation diagnosis method in any other appropriate manner (for example, by means of firmware).

[0158] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0159] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0160] As used in this disclosure, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and / or device (e.g., a magnetic disk, an optical disk, a memory, a programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0161] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0162] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0163] Computer systems may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The client and server relationship arises through computer programs running on the respective computers and having a client-server relationship to each other.

[0164] The foregoing description is intended only to provide specific embodiments of the present disclosure, intended to enable those skilled in the art to understand and implement the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the embodiments described herein, but rather to be construed in the broadest manner consistent with the principles and novel features disclosed herein.

Claims

1. A learning situation diagnosis method, characterized in that: include: Acquiring test question information of a diagnostic test question and answer data of a target subject for the diagnostic test question, wherein the test question information includes the difficulty level of a first test question; Generate a knowledge point difficulty level matrix for the target knowledge point corresponding to the diagnostic test question based on the test question information and the answer data, including: mapping the first test question difficulty to the second test question difficulty according to a first mapping rule; mapping the diagnostic test question score in the answer data to a score level based on a first preset threshold; mapping the score level to a third test question difficulty according to a mapping function corresponding to a second mapping rule, wherein the mapping function is constructed based on the test question difficulty threshold, difficulty slope, score level, and median of all test question scores; obtain a test question difficulty level based on the second test question difficulty and the third test question difficulty; and construct a knowledge point difficulty level matrix for the target knowledge point corresponding to the diagnostic test question based on the test question difficulty level; Based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, the target subject's mastery of the target knowledge point is predicted through the constructed level response model to obtain the knowledge point ability value, wherein the knowledge point discrimination matrix is ​​determined according to the diagnostic test questions.

2. The method according to claim 1, characterized in that The step of predicting the target subject's mastery of the target knowledge point by constructing a grade response model based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix to obtain the knowledge point ability value includes: Obtaining the simulation capability value and weight value corresponding to the target object; According to the simulated ability value, the knowledge point difficulty level matrix and the knowledge point discrimination matrix, the target subject's mastery of the target knowledge point is predicted by a constructed level response model, and the probability of the target knowledge point being associated with the score level is obtained; The knowledge point ability value is obtained according to the probability of the target knowledge point being associated with the score level, the simulated ability value and the weight value.

3. The method according to claim 1, characterized in that After obtaining the knowledge point capability value, the method further includes: Determine the knowledge point inspection frequency corresponding to the subject to be inspected where the target knowledge point is located; According to the knowledge point ability value and the knowledge point examination frequency, the subject ability value of the target object for the subject to be examined is determined through the constructed scoring model.

4. The method according to claim 3, characterized in that The step of determining the subject ability value of the target subject for the subject to be examined by constructing a scoring model based on the knowledge point ability value and the knowledge point examination frequency includes: Determining a candidate knowledge point inspection frequency from the knowledge point inspection frequencies; Performing smoothing and linear scaling on the normalized knowledge point ability values; According to the examination frequency of the candidate knowledge points and the processed knowledge point ability values, the subject ability value of the target object for the subject to be examined is determined.

5. The method according to claim 3, characterized in that After determining the subject ability value of the target subject for the subject to be examined, the method further includes: The comprehensive ability value of the target object for at least one subject to be examined is determined based on the subject ability value and the obtained subject weight value.

6. A learning situation diagnosis device, characterized in that: include: an acquisition module, configured to acquire test question information of a diagnostic test question and answer data of a target subject for the diagnostic test question, wherein the test question information includes the difficulty level of the first test question; A generation module, configured to generate a knowledge point difficulty level matrix of target knowledge points corresponding to the diagnostic test questions based on the test question information and the answer data; a prediction module, configured to predict the target subject's mastery of the target knowledge point by using a constructed grade response model based on the knowledge point difficulty level matrix and the knowledge point discrimination matrix, to obtain a knowledge point ability value; Wherein, the generating module is used for: Map the first test question difficulty to the second test question difficulty according to the first mapping rule; map the diagnostic test question score in the answer data to a score level based on a first preset threshold; map the score level to the third test question difficulty according to the mapping function corresponding to the second mapping rule, wherein the mapping function is constructed based on the test question difficulty threshold, difficulty slope, score level and the median of all test question scores; obtain the test question difficulty level according to the second test question difficulty and the third test question difficulty; and construct a knowledge point difficulty level matrix for the target knowledge point corresponding to the diagnostic test question based on the test question difficulty level.

7. An electronic device, characterized in that: The electronic device comprises: processor; and Memory for storing programs, Wherein, the program includes instructions, and when the instructions are executed by the processor, the processor executes the learning situation diagnosis method according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable the computer to execute the learning situation diagnosis method according to any one of claims 1 to 5.

Images