Learning situation analysis method and related device, equipment and storage medium

Abstract

The application discloses a learning situation analysis method and related devices, equipment and storage media, wherein the learning situation analysis method comprises: obtaining an evaluation result and a target difficulty score of a subject; wherein the evaluation result at least comprises a grade score representing the learning level of the subject; and generating a learning situation analysis result of the subject based on the evaluation result and the difficulty score. The above scheme can improve the accuracy of learning situation analysis of the subject.

Description

[0001] This application claims priority to Chinese patent application filed on December 19, 2023, with application number 2023117594646 and title "Learning Analysis Method and Related Devices, Equipment and Storage Medium", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of intelligent education technology, and in particular to a learning analysis method and related devices, equipment and storage media. Background Technology

[0003] With the continuous development of intelligent education, student evaluation, as an important part of educational evaluation, can play a positive guiding and motivating role in improving students' academic performance and overall quality.

[0004] In existing technologies, learning ability is typically assessed based on the average accuracy rate of subjects' recent answer records. However, because the difficulty of the questions chosen by subjects varies and the coverage of the answer records is somewhat limited, the accuracy of learning analysis is low. Therefore, improving the accuracy of learning analysis for subjects has become an urgent problem to be solved. Summary of the Invention

[0005] The main technical problem addressed by this application is to provide a learning analysis method and related devices, equipment, and storage media that can improve the accuracy of learning analysis of test subjects.

[0006] To address the aforementioned technical issues, the first aspect of this application provides a learning analysis method, comprising: obtaining the assessment results of the test subjects and the difficulty scores of the target questions; wherein the assessment results include at least a grade score representing the learning level of the test subjects; and generating learning analysis results of the test subjects based on the assessment results and the difficulty scores.

[0007] To address the aforementioned technical problems, a second aspect of this application provides a learning analysis device, comprising: an acquisition module and a generation module. The acquisition module is used to acquire the assessment results of the test subjects and the difficulty scores of the target questions; wherein the assessment results include at least a grade score representing the learning level of the test subjects; and the generation module is used to generate learning analysis results of the test subjects based on the assessment results and the difficulty scores.

[0008] To address the aforementioned technical problems, a third aspect of this application provides an electronic device, including a memory and a processor coupled to each other. The memory stores program instructions, and the processor executes the program instructions to implement the learning analysis method described in the first aspect.

[0009] To address the aforementioned technical problems, the fourth aspect of this application provides a computer-readable storage medium storing program instructions executable by a processor, the program instructions being used to implement the learning analysis method of the first aspect.

[0010] The above scheme generates learning analysis results for the test subjects based on their assessment results and the difficulty scores of the target questions. It uses both the assessment results, which at least represent the test subjects' learning level, and the difficulty scores, which represent the difficulty of the target questions, as reference information for generating the learning analysis results. This allows for learning analysis based on multi-dimensional data, thus improving the accuracy of learning analysis for the test subjects. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating one embodiment of the learning analysis method of this application; Figure 2 This is a flowchart illustrating another embodiment of the learning analysis method in this application; Figure 3 This is a schematic diagram of the framework of an embodiment of the learning evaluation comment generation model of the learning analysis method in this application; Figure 4 This is a flowchart illustrating yet another embodiment of the learning analysis method proposed in this application; Figure 5 This is a schematic diagram of an embodiment of the target grade distribution of the subjects in the learning analysis method of this application; Figure 6 This is a schematic diagram of the framework of an embodiment of the learning analysis device of this application; Figure 7 This is a schematic diagram of the framework of an embodiment of the electronic device of this application; Figure 8 This is a schematic diagram of a framework of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0012] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0013] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0014] In this paper, the terms "system" and "network" are often used interchangeably. The term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the slash " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this paper means two or more.

[0015] Please see Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the learning analysis method of this application. Specifically, it may include the following steps: Step S10: Obtain the test results of the subjects and the difficulty score of the target questions.

[0016] In this embodiment of the disclosure, the evaluation results include at least a rating scale that characterizes the learning level of the test subject.

[0017] In one implementation scenario, the rating scale represents the learning ability of the test subjects. Specifically, the higher the rating scale score, the better the learning ability of the corresponding test subject, and the lower the rating scale score, the worse the learning ability of the test subject. It should be noted that the upper and lower limits of the test subject's rating scale score are not limited in this application.

[0018] In one implementation scenario, when subjects answer the target questions for the first time within a certain preset period, they are given an initial rating score. It is understood that the initial rating score can be set based on those skilled in the art, or it can be dynamically set based on the average rating score of all subjects in the previous preset period, or it can be adjusted based on the final rating score at the end of the previous preset period. Examples will not be given here.

[0019] In a specific implementation scenario, the preset period can be determined based on the learning objectives of the test subjects. For example, the preset period can be one semester or three months, and so on.

[0020] Step S20: Based on the assessment results and difficulty scores, generate the learning analysis results of the test subjects.

[0021] In this embodiment of the disclosure, the learning analysis results are used to characterize the learning level and problem-solving level of the test subjects. Specifically, the learning analysis results are presented in text form, numerical form, etc., which are not limited in this application.

[0022] In one implementation scenario, as a possible approach, a learning analysis model can be pre-trained. This model can include, but is not limited to, an Encoder-Decoder architecture network model. The evaluation results and difficulty scores are input into the learning analysis model, and its output is used as the learning analysis result for the test subjects. To ensure the model's prediction accuracy as much as possible, sample data pairs of several test subjects can be collected. Each data pair includes the sample evaluation results, sample difficulty scores, and the actual analysis results labeled on the data pairs. Based on this, the learning analysis model can process the data pairs to obtain the sample predicted analysis results for the test subjects. Then, based on the difference between the actual and predicted analysis results, the network parameters of the learning analysis model can be adjusted until the model training converges. Finally, the converged learning analysis model can be used to process the test subjects' evaluation results and the difficulty scores of the target questions to obtain the test subjects' learning analysis results. It should be noted that the specific processing procedures of the learning analysis model can be found in technical details such as those of Encoder-Decoder architecture network models, and will not be elaborated upon here.

[0023] The above scheme generates learning analysis results for the test subjects based on their assessment results and the difficulty scores of the target questions. It uses both the assessment results, which at least represent the test subjects' learning level, and the difficulty scores, which represent the difficulty of the target questions, as reference information for generating the learning analysis results. This allows for learning analysis based on multi-dimensional data, thus improving the accuracy of learning analysis for the test subjects.

[0024] Please see Figure 2 , Figure 2 This is a flowchart illustrating another embodiment of the learning analysis method proposed in this application. Specifically, it may include the following steps: Step S11: Obtain the test results of the subjects and the difficulty score of the target questions.

[0025] In this embodiment of the disclosure, the learning analysis results include at least learning comments about the test subjects. Specifically, the learning comments are presented in natural language. Natural language is based on the natural and informal language system used by humans and can be used for communication, expressing opinions, conveying information, etc. It should be noted that the language type of the learning comments is not limited in this application, such as Chinese, English, French, etc.

[0026] In one implementation scenario, the target knowledge points contained in the target question are obtained, along with the historical learning data and historical problem-solving data of the test subjects. Based on the historical learning data, the historical learning progress of the test subjects regarding the target knowledge points is determined, and based on the historical problem-solving data, the historical problem-solving accuracy rate of the test subjects regarding the target knowledge points is determined. For example, if the target question is "Given that a, b, and c are three consecutive columns of an arithmetic sequence, and a + b + c = 21, what is the value of b?", the knowledge point contained in the question is "arithmetic sequence". The historical learning progress of the test subjects regarding the target knowledge point can be determined based on their learning progress in courses related to "arithmetic sequence". There are four lessons on "arithmetic sequences." The subjects have completed the first two lessons, resulting in a historical learning progress score of 0.5 for the target knowledge point. Based on this historical learning progress and historical problem-solving accuracy, a grade score is generated for each subject. Both historical learning progress and historical problem-solving accuracy are directly proportional to the grade score. This means that a higher learning progress indicates a higher level of completion of the knowledge point, and a higher problem-solving accuracy indicates a higher level of learning. The above approach, based on the subjects' historical learning and problem-solving data, provides more targeted historical learning progress and accuracy data for the target knowledge point, generating a grade score for that knowledge point. Therefore, it improves the accuracy of learning analysis for the subjects.

[0027] In a specific implementation scenario, the test subjects' historical test-taking data includes their answers to questions covering various knowledge points. After obtaining the target knowledge points contained in the target questions, the historical test-taking data is filtered based on the target knowledge points. Questions in the test-taking data containing the target knowledge points are selected as candidate questions. Then, the historical test-taking accuracy rate used to generate the grade score is obtained based on the accuracy rate of the candidate questions. Specifically, the candidate questions can contain only the target knowledge points, or they can contain the target knowledge points and other knowledge points. For example, "Given that a, b, and c are three consecutive columns of an arithmetic sequence, and a + b + c = 21, what are the factors of b?", which simultaneously contains the two knowledge points of "arithmetic sequence" and "factors".

[0028] In a specific implementation scenario, the subject's grade score ranges from 0 to 1, with corresponding weights assigned to historical learning progress and historical test-taking accuracy, respectively, to calculate the grade score. For example, if the subject's historical learning progress is 0.5 and historical test-taking accuracy is 0.8, and the weight of historical learning progress is set to be equal to the weight of historical test-taking accuracy, both being 0.5, then the subject's grade score is calculated to be 0.65. It should be noted that the above embodiment is only one possible implementation method, and the method of calculating the grade score is not limited in this application.

[0029] In one implementation scenario, the evaluation results also include data on the test subjects' answers to the target questions and data on the process of answering the questions. For example, the answer data includes any one of answering all questions correctly, answering some questions correctly, or answering all questions incorrectly. It can be understood that when the standard answer to the target question includes several solution steps, or when the target question has multiple unanswered questions, the answer data is any one of answering all questions correctly, answering some questions correctly, or answering all questions incorrectly. In some implementation scenarios, the answer data also includes all the test subjects' answers to the target question.

[0030] In a specific implementation scenario, the data on the test-taking process includes the time it takes for the test subject to answer the target question and the relevant analytical materials viewed during the problem-solving process. For example, during the process of the test subject answering the target question, relevant knowledge point videos, knowledge point cards, interactive analysis of question knowledge points, etc. are provided to the test subject. The test subject can generate corresponding test-taking process data by viewing the relevant content.

[0031] In a specific implementation scenario, data on the question-answering process can be used to measure the level of seriousness of the test subjects in answering the target questions, such as determining whether there are instances of casual or rapid answering, or behavioral data such as previewing answers.

[0032] In a specific implementation scenario, data on the test-taking process can be used to measure the completion rate of the test subjects' learning process, such as determining whether they watched knowledge point videos or knowledge cards, or viewed interactive explanations.

[0033] In one implementation scenario, the target test point information contained in the target question is obtained, and the target test point information includes at least one of the following: knowledge graph of the target test point, difficulty level of the target test point, and question distribution of the target test point. The difficulty score of the target question is generated based on the target test point information.

[0034] In a specific implementation scenario, the knowledge graph of the target test point in the target test point information is used to describe the sequential relationship between different test point knowledge. For example, test point A is a precursor test point to test point B, meaning that mastering test point A helps to better master test point B. Another example is that each knowledge point can be divided into one of the following categories based on the knowledge graph of that knowledge point: "basic knowledge point," "advanced knowledge point," or "advanced knowledge point."

[0035] In a specific implementation scenario, the distribution of test questions for the target test points includes attribute information such as test frequency, distribution of actual test scores, and distribution of actual test difficulty.

[0036] In a specific implementation scenario, the difficulty score of the target question and the level score of the test subject are set to the same range of values. For example, the level score of the test subject is set to a range of 0 to 1, and the difficulty score of the target question is also set to a range of 0 to 1.

[0037] Step S21: Based on the evaluation results and difficulty score, select the target feedback statement and the target learning statement.

[0038] In one implementation scenario, when the evaluation results also include the test subject's answering results and answering process data for the target question, candidate feedback statements are selected as the target feedback statements based on the difficulty score of the target question and the test subject's answering results and level scores. The candidate feedback statements include several encouraging statements representing different levels of feedback and several critical statements representing different levels of feedback. For example, candidate feedback statements include a first-level encouraging statement "Keep it up!", a second-level encouraging statement "You're great, you answered this question correctly!", and a third-level encouraging statement "Fantastic! You solved a difficult problem!". Candidate feedback statements also include a first-level critical statement "Please carefully understand the content provided in the question stem!" and a second-level critical statement "Please don't give a perfunctory answer." It is understood that the above embodiments are only one possible implementation method, and candidate feedback statements representing the same level of feedback may use different expressions. The specific content and number of candidate feedback statements are not limited in this application. Furthermore, based on the difficulty score of the target question and the test subject's question-solving process data and level score, candidate learning statements are selected as target learning statements. Candidate learning statements include at least one of question-solving evaluation statements and learning suggestion statements. For example, question-solving evaluation statements include "Step 5 used an incorrect calculation formula, resulting in an incorrect final calculation result," and learning suggestion statements include "Please watch the instructional video a about the formula to consolidate your learning content," etc. It is understood that the above embodiments are only one possible implementation method, and the specific content of the target learning statement is not limited in this application.

[0039] In a specific implementation scenario, candidate feedback statements are labeled with corresponding feedback level tags. For example, "You did a great job" and "You're amazing" are both labeled with the feedback level tag pair "Level 2 - Encouragement", and "You were too careless when doing the questions" is labeled with the feedback level tag pair "Level 2 - Criticism", etc.

[0040] In a specific implementation scenario, the learning suggestion statements in the candidate learning statements are associated with relevant knowledge points involved in the question-solving process data. For example, if a "basic knowledge point" is detected in the question-solving process data, the learning suggestion statement can recommend reinforcement questions for this "basic knowledge point" or learning videos for "advanced knowledge points" related to this "basic knowledge point." Alternatively, when the subject fails to answer the target question correctly, the learning suggestion statement can recommend learning videos, learning cards, etc., related to this "basic knowledge point."

[0041] In a specific implementation scenario, a target baseline value is determined based on whether the test subjects correctly answered the target questions, as represented by the test result data. For example, if all test subjects answered the target questions correctly, the target baseline value is set to the maximum value of 1. If some test subjects answered the target questions correctly, the target baseline value is determined according to the proportion of the test subjects' correct answers. For example, if the target questions include question 1 and question 2, and the test subjects answered question 1 correctly but answered question 2 incorrectly, the target baseline value is set to 0.5. As another example, if the answer to the target question includes 5 complete steps, and the test subjects answered 3 complete steps correctly, the target baseline value is set to 0.6. If all test subjects answered the target questions incorrectly, the target baseline value is set to the minimum value of 0.1. It should be noted that the range of the target baseline value is not limited in this application. The first weight is determined based on the difficulty score, and the second weight is determined based on the level score. The first weight is directly proportional to the difficulty score, and the second weight is inversely proportional to the level score. The target benchmark value is weighted based on the first weight and the second weight respectively to obtain a weighted sum. For example, the level score of the test subject ranges from 0 to 1, and the difficulty score of the target question also ranges from 0 to 1. The higher the level score, the stronger the test subject's learning ability. The higher the difficulty score, the more difficult the target question. Candidate feedback statements that correspond to the target feedback level with the sum value are selected as the target feedback statements.

[0042] In a specific implementation scenario, a mapping relationship is set between the difficulty score and the first weight. For example, when the difficulty score is greater than 0.8 and not greater than 1, the mapped first weight is 1.5; when the difficulty score is greater than 0.6 and not greater than 0.8, the mapped first weight is 1.2; when the difficulty score is greater than 0.4 and not greater than 0.6, the mapped first weight is 1; when the difficulty score is greater than 0.2 and not greater than 0.4, the mapped first weight is 0.8; and when the difficulty score is greater than 0 and not greater than 0.2, the mapped first weight is 0.6.

[0043] In a specific implementation scenario, a mapping relationship is set between the grade score and the second weight. For example, when the grade score is greater than 0.8 and not greater than 1, the mapped second weight is 0.6; when the grade score is greater than 0.6 and not greater than 0.8, the mapped second weight is 0.8; when the grade score is greater than 0.4 and not greater than 0.6, the mapped second weight is 1; when the grade score is greater than 0.2 and not greater than 0.4, the mapped second weight is 1.2; and when the grade score is greater than 0 and not greater than 0.2, the mapped second weight is 1.5.

[0044] In a specific implementation scenario, a mapping relationship is set between different values ​​and candidate feedback statements with different feedback levels. After calculating the sum, the corresponding candidate feedback statement is selected as the target feedback statement based on the mapping relationship. For example, taking the aforementioned embodiment as an example, the target baseline value is 0.5, the difficulty of the target question is 0.7 (i.e., the first weight is 1.2), the subject's level is 0.7 (i.e., the second weight is 0.8), and the calculated sum is 1. The feedback level label pair corresponding to the candidate feedback statement with the value 1 can be obtained as "Level 1 - Encouragement". Accordingly, if the calculated sum is less than the preset encouragement threshold, such as 0.3, the selected candidate feedback statement is a criticism statement. When the calculated sum is not less than the preset encouragement threshold, the selected candidate feedback statement is an encouragement statement, and the feedback level changes accordingly with the change of the sum value.

[0045] It should be noted that the above embodiments are only one possible implementation method, and the methods for generating the first weight and the second weight, as well as the methods for selecting the target feedback statement, are not limited in this application.

[0046] Step S31: Generate learning progress comments based on the target feedback statement and the target learning statement.

[0047] In one implementation scenario, the learning evaluation comments are presented in natural language. Natural language is based on the natural and informal language system used by humans and can be used for communication, expressing opinions, conveying information, etc. It should be noted that the language type of the learning evaluation comments is not limited in this application, such as Chinese, English, French, etc.

[0048] In a specific implementation scenario, the output format of learning evaluation comments is not limited in this application, such as text, voice, images, etc.

[0049] In a specific implementation scenario, the learning evaluation comments also include knowledge content related to the target question, such as learning videos and extended exercises.

[0050] In one implementation scenario, based on the feedback level label pair contained in the target feedback statement and the knowledge point content contained in the target learning statement, a learning evaluation comment is generated after polishing. For example, the target feedback statement is "Please keep up the good work," and the feedback level label pair is "Level 1 - Encouragement." The target learning statement is "Please continue to study the relevant courses on arithmetic sequences," and the knowledge point content is "arithmetic sequences." The refined learning evaluation comment is "Your knowledge of arithmetic sequences is a bit weak. Please keep up the good work. It is recommended that you watch the video course on arithmetic sequences to learn about arithmetic sequences."

[0051] The above scheme, based on the test results of the subjects and the difficulty score of the target questions, selects target feedback statements and target learning statements. The target feedback statements include either encouraging or critical statements, while the target learning statements instruct the subjects on the learning content. Based on these statements, learning progress comments are generated. The scheme uses both the test results, which at least represent the subjects' learning level, and the difficulty score, which represents the difficulty of the target questions, as reference information for generating these comments. This allows for learning progress analysis based on multi-dimensional data, avoiding bias caused by judging the subjects' learning level solely based on whether they answered the target questions correctly. Furthermore, the learning progress comments, including both target feedback and target learning statements, can enhance the subjects' enthusiasm for learning, provide timely feedback on their learning progress and offer corresponding instructions, thereby improving learning outcomes. Therefore, it can improve the accuracy of learning progress analysis for the subjects.

[0052] Please see Figure 3 , Figure 3 This is a schematic diagram of the framework of an embodiment of the learning performance evaluation generation model 30 in the learning performance analysis method of this application. For example... Figure 3 As shown, the learning evaluation generation model 30 includes a difficulty score generation network 31, a grade score generation network 32, a sentence generation network 33, and a large language model 34. Based on the target feedback sentence and the target learning sentence, it constructs prompt text. The prompt text is used to indicate the evaluation generation style of the large language model 34 and to adjust the knowledge text about the learning direction of the test subject. The prompt text is sent to the large language model 34, and the target sentence output by the large language model 34 serves as the learning evaluation of the test subject.

[0053] In one implementation scenario, the Large Language Model 34 can understand human natural language input and, based on the input text, obtain keyword information and logical relationships between various elements, generating semantically relevant output. For more technical details on the specific principles, please refer to the Large Language Model documentation; they will not be elaborated upon here. For example, if the target feedback statement is "Great," and the target learning statement is "High mastery of key points and high accuracy in answering questions," the constructed prompt text could be something like, "Please combine 'Great' with an attempt to refine the tone and style of the learning feedback, and generate a learning feedback based on the specific content of the current target learning statement." The generated learning feedback could be something like, "Dear student, you have a thorough grasp of the key points in this unit, and your answer accuracy is very high. This shows that your foundation is very solid, which is commendable! We suggest you continue learning advanced key points related to the current key points." This leverages the general understanding capabilities of the Large Language Model 34 to generate the learning feedback. It should be noted that the Large Language Model 34 can include, but is not limited to, open-source large models such as LLAMA and Bloom; the network architecture of the Large Language Model is not limited here.

[0054] The above scheme, based on the test results of the subjects and the difficulty score of the target questions, selects target feedback statements and target learning statements. The target feedback statements include either encouraging or critical statements, while the target learning statements instruct the subjects on the learning content. Based on these statements, learning progress comments are generated. The test results, which at least represent the subjects' learning level, and the difficulty score, which represents the difficulty of the target questions, are used together as reference information for generating the comments. This allows for learning progress analysis based on multi-dimensional data, avoiding bias caused by judging the subjects' learning level solely based on whether they answered the target questions correctly. Furthermore, learning progress comments including target feedback and target learning statements can enhance the subjects' enthusiasm for learning, provide timely feedback on their learning progress and offer corresponding instructions, thus improving learning outcomes. Learning progress comments refined based on a large language model can further enhance the subjects' enthusiasm. Therefore, it can improve the accuracy of learning progress analysis for the subjects.

[0055] Please see Figure 4 , Figure 4 This is a flowchart illustrating another embodiment of this application. Specifically, it may include the following steps: Step S12: Obtain the test results of the test subjects and the difficulty score of the target questions.

[0056] In this embodiment of the disclosure, the learning analysis results include at least the updated grade score of the test subject and the updated difficulty score of the target question.

[0057] In one implementation scenario, the same target question can be selected and answered by multiple subjects. When the target question is not selected and answered by any subject, an initial difficulty score is set. For example, the initial difficulty score can be set by those skilled in the art or determined based on the average difficulty score of all questions in the question bank. Examples will not be given here.

[0058] In a specific implementation scenario, several questions are stored in a question bank. The test subjects select a question from the question bank as the target question. The difficulty score of each question in the question bank is dynamically adjusted based on the selected answers of several test subjects. It can be understood that when a question is selected as the target question by a sufficient number of test subjects and answered, the difficulty score of the target question will stabilize within a certain range of difficulty scores. This allows the difficulty score of each question to accurately represent the average difficulty, improves the accuracy of the question difficulty score, and thus improves the accuracy of the learning analysis of the test subjects.

[0059] In a specific implementation scenario, in response to the subject selecting any question from the question bank, the selected question is determined as the target question, and the subject's latest level score and the difficulty score of the target question are obtained. It should be noted that the method by which the subject selects the target question is not limited in this application.

[0060] Step S22: Based on the subject's level score and the difficulty score of the target question, obtain the predicted score rate of the subject's answer to the target question, and obtain the subject's actual score rate of answering the target question.

[0061] In one implementation scenario, the rating and difficulty scores are derived from the Elo rating system, an evaluation method for measuring the level of various game-related activities. In this scheme, the test-taker and the target problem are treated as a pair of game-takers, and the test-taker's ability to answer questions is a normally distributed random variable. For example, the test-taker's ability to answer questions is influenced by factors such as knowledge reserves and answering status, thus exhibiting a normal distribution. The Elo rating system uses the average of the random variable to represent the test-taker's true level; that is, the rating score characterizes the test-taker's learning ability. Similarly, the difficulty of the target problem is a normally distributed random variable, and the Elo rating system uses the average of the random variable to represent the true difficulty of the target problem; that is, the rating score characterizes the difficulty of the target problem. This scheme, based on the Elo rating system, treats the test-taker and the target problem as a pair of game-takers, dynamically adjusting the test-taker's rating and the target problem's difficulty score to improve the accuracy of the test-taker's learning analysis.

[0062] In a specific implementation scenario, the level scores of the test subjects and the difficulty scores of the target questions use the same scoring criteria. For example, the level score ranges from 0 to 100, and the difficulty score ranges from 0 to 100 as well. Furthermore, the distribution trend of the level scores of all test subjects is similar to the distribution trend of the difficulty scores of all target questions.

[0063] In one implementation scenario, based on the subject's rating score and the difficulty score of the target question, the predicted score rate of the subject's answer to the target question can be obtained. Understandably, the Elo rating system can make predictions based on the subject's rating score and the difficulty score of the target question to obtain the predicted score rate of the subject's answer to the target question. When the subject's rating score is greater than the difficulty score of the target question, the predicted score rate is higher, indicating that the subject is more likely to answer the target question correctly. When the subject's rating score is not greater than the difficulty score of the target question, the predicted score rate is lower, indicating that the subject is more likely to answer the question incorrectly. Moreover, the magnitude of the predicted score rate is also related to the magnitude of the difference between the rating score and the difficulty score.

[0064] In a specific implementation scenario, the predicted score rate can be predicted based on the target question's set score. For example, if a target question is worth 10 points, based on the subject's level score and the target question's difficulty score, it is predicted that the subject can score 6 points on the target question, thus determining the predicted score rate for the subject's answer to this target question to be 0.6. It should be noted that the above embodiment is only one possible implementation method, and for the sake of brevity, it will not be described in detail here.

[0065] In a specific implementation scenario, the Elo rating system has a target parameter to characterize the test-taker's answering ability. It obtains the first difference between the difficulty score and the rating score, and then obtains the target parameter determined for the test-taker based on the Elo rating system's distribution interval. Based on the first difference and the target parameter, the predicted score rate is obtained. See the following formula for details:

[0066] In formula (1), Characterizing the rating scores of the test subjects, The difficulty score represents the target question. Characterizing the target parameters. It should be noted that the specific values ​​of the target parameters involved in the above formula are only one possible implementation method and are not limited in this application.

[0067] In one implementation scenario, the true score rate of the subject's answers to the target question is obtained. For example, the true score rate of the subject's correct answers to the target question is 1, and the true score rate of the subject's incorrect answers to the target question is 0. Alternatively, if the target question is worth 10 points, and the subject answers the target question and gets 6 points, the subject's true score rate for answering this target question is determined to be 0.6. It should be noted that the above embodiments are only one possible implementation method, and for the sake of brevity, they will not be described in detail here.

[0068] In a specific implementation scenario, after the subject confirms their answer, the true score rate is obtained based on the subject's answer, and the predicted score rate is not affected by the result of this answer.

[0069] Step S32: Update at least one of the rating score and difficulty score based on the predicted score rate and the actual score rate.

[0070] In one implementation scenario, after obtaining the predicted score rate and the true score rate, and before updating the grade score, a first adjustment coefficient is obtained to represent the score update step size. For example, the first adjustment coefficient is 64, 32, etc., and will not be listed one by one here.

[0071] In a specific implementation scenario, after obtaining the first adjustment coefficient, a first adjusted score is determined based on the predicted score rate, the actual score rate, and the first adjustment coefficient. The grade score is then updated based on this first adjusted score. In this scheme, the subject's performance on each target question is reflected in the changes in the grade score. Furthermore, the grade score is dynamically updated based on the varying difficulty levels of each target question, thus improving the accuracy of learning analysis for the subjects.

[0072] In a specific implementation scenario, based on the total number of questions answered by the subject within a preset period, a first cumulative number of questions answered by the subject during the preset period is determined, and a first adjustment coefficient is determined based on this first cumulative number of questions. The preset period can be referred to in the detailed description of the foregoing embodiments, and will not be repeated here for brevity. The above scheme, by determining the first adjustment coefficient based on the first cumulative number of questions answered by the subject within the preset period, can dynamically adjust the update pace of the grade scores, thereby improving the accuracy of learning analysis.

[0073] In a specific implementation scenario, a formula for determining the first adjustment coefficient is established based on whether the first cumulative number of questions reaches a preset question threshold. For example, to enable test subjects to quickly update their level scores to the average level of learning ability in the early stages of a preset period, a preset question threshold of 32 questions is set. When the first cumulative number of questions is no more than 32, the first adjustment coefficient is calculated using the following formula:

[0074] In formula (2), Characterizing the first adjustment factor, This represents the current question number. When the first cumulative number of questions exceeds 32, the first adjustment coefficient is set to 16. The above scheme determines the first adjustment coefficient based on the range of the first cumulative number of questions. When the cumulative number of questions completed by the subject does not exceed the preset question threshold, the subject's grade level can be quickly adjusted to near the average learning ability level in the early stages of the preset period. When the cumulative number of questions completed by the subject exceeds the preset question threshold, to avoid the subject excessively completing questions to improve their grade level, the first adjustment coefficient is set to an appropriate value to ensure the accuracy of the learning analysis.

[0075] In a specific implementation scenario, the second difference between the true score and the predicted score is obtained. The first adjusted score is then calculated by multiplying this second difference by the first adjustment coefficient. See the following formula for details:

[0076] In formula (3), The updated rating scores of the subjects represent the new ratings. The ratings of the subjects before the update were represented. Characterizing the first adjustment factor, Characterizing the true score rate, Characterization prediction score rate The first adjustment score is represented by the above scheme. The subject's performance on each target question is reflected in the change of their grade score. Furthermore, the grade score is dynamically updated based on the varying difficulty of each target question, thus improving the accuracy of learning analysis. When the difficulty score of a target question answered by the subject is greater than their current grade score, the reward for a correct answer is larger (a larger increase in grade score), while the penalty for an incorrect answer is smaller (a smaller decrease in grade score). Conversely, when the difficulty score of a target question answered by the subject is less than their current grade score, the reward for a correct answer is smaller (a smaller increase in grade score), while the penalty for an incorrect answer is larger (a larger decrease in grade score). This encourages subjects to continuously challenge themselves with questions slightly above their current learning ability, thereby improving their skills and engaging in learning more actively.

[0077] In a specific implementation scenario, the first adjustment coefficient is determined based on the total number of questions answered by the subject within a preset period. The preset period includes several sub-periods; for example, a preset period might be a semester, with each complete week from Monday to the weekend constituting a sub-period. The first adjustment score is also related to the number of questions answered by the subject in each sub-period. The sub-period to which the subject currently answers the target question is located is determined as the target sub-period. Based on the total number of questions answered within the target sub-period, the second cumulative number of questions answered by the subject within the target sub-period is determined. Based on the grade score and the second cumulative number of questions, the first preset coefficient is determined, and this first preset coefficient represents the adjustment step of the first adjustment score. The first adjustment score is determined based on the product of the second difference, the first adjustment coefficient, and the first preset coefficient. For example, if one day is a sub-period, the first preset coefficient is 1.15 if the number of questions answered that day is no more than 8; it is 1 if the number is more than 8 and no less than 16; and it is 0.85 if the number is more than 16. The above approach helps ensure that participants answer the target questions in each sub-cycle, avoiding occasional cramming and thus cultivating good study habits.

[0078] In a specific implementation scenario, the target level of the subject is determined based on the mapping relationship between the subject's rating score and the level set. The level set contains several candidate levels, as detailed in Table 1 below:

[0079] Table 1 As shown in Table 1, based on the range of ability scores, the learning ability of the subjects is represented by the target level and sub-segment, which makes the display of learning ability more concrete, so as to encourage the subjects to actively update the target level and improve the learning effect of the subjects.

[0080] Please see Figure 5 , Figure 5 This is a schematic diagram of an embodiment of the target level distribution of the subjects in the learning analysis method of this application. For example... Figure 5 As shown, the target level distribution of the test subjects exhibits a normal distribution trend, meaning that the test subjects' answers to each target question are reflected in the changes in their level scores. Furthermore, the level scores of the test subjects are dynamically updated based on the different levels of difficulty of each target question, thus improving the accuracy of the analysis of the test subjects' learning situation.

[0081] In a specific implementation scenario, a second preset coefficient is determined based on the target level currently held by the subject. The second preset coefficient represents the adjustment step of the first adjustment score. The first adjustment score is determined based on the product of the second difference, the first adjustment coefficient, and the second preset coefficient.

[0082] In a specific implementation scenario, a second preset coefficient is matched based on the subjects' target levels. For example, referring to Table 1, the second preset coefficient is 1.15 for subjects with target levels "D" and "E", 1.05 for subjects with target levels "B" and "C", 0.95 for subjects with target levels "A", and 0.85 for subjects with target levels "S". This scheme adjusts the difficulty of improving the target level and score for subjects with weaker representational learning abilities, while increasing the difficulty for subjects with stronger representational learning abilities.

[0083] In a specific implementation scenario, before updating at least one of the grade score and difficulty score based on at least the predicted score rate and the actual score rate, a second adjustment coefficient is obtained. This second adjustment coefficient represents the update step size of the difficulty score. Based on the predicted score rate, the actual score rate, and the second adjustment coefficient, a second adjusted score is determined. Based on this second adjusted score, the difficulty score is updated. This scheme dynamically assesses the difficulty of each target question based on the participant's answers, continuously updating the difficulty score of each question. This ensures that the difficulty score of each question accurately represents the average difficulty, improving the accuracy of the question difficulty scores and thus enhancing the accuracy of learning analysis of the participants.

[0084] In a specific implementation scenario, the difficulty score is updated based on the second adjustment score, as detailed in the following formula:

[0085] In formula (4), The difficulty score represents the updated target question. The difficulty score of the target question before the update is represented. Characterizing the second adjustment coefficient, Characterizing the true score rate, Characterization prediction score rate The second adjustment score is used to represent the difficulty. The above scheme dynamically assesses the difficulty of each target question based on the subjects' answers, continuously updating the difficulty score of each question. This ensures that the difficulty score of each question accurately represents the average difficulty, improving the accuracy of the question difficulty score and thus enhancing the accuracy of the learning analysis of the subjects.

[0086] In a specific implementation scenario, based on the difficulty score of the target question, the corresponding difficulty level is matched. For example, the questions are divided into different difficulty levels such as "easy", "basic" and "advanced" to avoid the bias caused by manual labeling. Moreover, this detailed difficulty division helps to meet the learning needs of test subjects at different levels and improves learning efficiency.

[0087] In one implementation scenario, after updating at least one of the level score and difficulty score based on at least the predicted score rate and the actual score rate, the latest level score of the subject and the latest difficulty score of each question are obtained. Based on the latest level score, the range of difficulty scores for recommending questions to the subject is determined. Based on the range of difficulty scores and the latest difficulty scores of each question, recommended questions are determined. This scheme assesses the subject's learning ability based on the latest level score and determines the difficulty of each question based on the latest difficulty scores, thereby improving the accuracy of recommended questions and minimizing the possibility of recommending questions that are too difficult or too easy.

[0088] In a specific implementation scenario, a recommended difficulty threshold is set. For example, if the recommended difficulty threshold is 5, and the latest level score obtained from the test subject is 65, then the recommended difficulty score range for the questions is 60 to 70.

[0089] In a specific implementation scenario, recommended questions are displayed, and the display order is determined based on the difference between the difficulty score of the recommended question and the participant's level score. For example, recommended questions with a difficulty score greater than their level score are displayed at the front end, and then arranged on the recommended question display page in ascending order of score difference. This method, while encouraging participants to tackle more challenging target questions, aims to improve the quality of the recommended questions and actively promote participants' selection and response to them.

[0090] In a specific implementation scenario, after determining recommended questions based on the difficulty score range and the latest difficulty scores of each question, in response to the subject's instruction to select a recommended question, the system obtains the level score update coefficient, selects the recommended question as the new target question, and executes the steps of obtaining the subject's level score and the difficulty score of the target question, as well as subsequent steps, to update at least one of the subject's level score and the difficulty score of the recommended question. The updated level score is based on the level score update coefficient, the predicted score rate, and the actual score rate. This scheme incentivizes subjects to select recommended questions by using the level score update coefficient, which helps subjects perform personalized learning. Recommending better questions based on the subject's level score and the difficulty scores of each question improves the accuracy of learning analysis. Understandably, to incentivize subjects to select recommended questions, the level score update coefficient should be no less than 1 when a subject selects a recommended question.

[0091] In another specific implementation scenario, after determining recommended questions based on the difficulty score range and the latest difficulty score of each question, if the test subject does not trigger a selection instruction for the recommended questions, the level score update coefficient can be appropriately adjusted, provided that the adjusted level score update coefficient is not greater than 1. This approach avoids test subjects selecting target questions unsuitable for their learning abilities, thereby improving the efficiency of learning analysis.

[0092] In a specific implementation scenario, a historical answer database is built based on all the questions the subject has already answered. Before obtaining the subject's grade score and the difficulty score of the target question, it checks whether the currently selected target question is the same as a question in the historical answer database. If a previously answered question with the same target question exists in the historical answer database, a pop-up window or other means is used to remind the subject that they have already answered the question and whether they want to continue answering. If the subject chooses to continue answering, the actual score rate of the current answer is compared with the actual score rate of the previous answer. If the actual score rate of the current answer is greater than the actual score rate of the previous answer, the subject's grade score is increased. If the actual score rate of the current answer is equal to the actual score rate of the previous answer, the subject's grade score remains unchanged. If the actual score rate of the current answer is less than the actual score rate of the previous answer, the subject's grade score is decreased. Moreover, the increase in the subject's grade score is no greater than the decrease, to prevent the subject from doing too many practice questions and to encourage the subject to improve their memory of questions they have already practiced.

[0093] In one implementation scenario, after updating at least one of the level score and difficulty score based on at least the predicted score rate and the actual score rate, a trend graph of the subject's level score change is generated and displayed based on the level scores updated after each round of answering questions. This method allows subjects to view the trend graph representing changes in their learning ability in real time. When the trend graph shows a gradual improvement in their learning ability, it generates a sense of accomplishment and satisfaction, increasing motivation. When the trend graph shows a gradual decline in their learning ability, it encourages them to actively adjust their learning attitude, helping them learn more efficiently. The level score feedback enhances the user experience, increasing user satisfaction and loyalty. By continuously tracking the subject's answering behavior, the trend graph forms an intuitive curve of level score changes. This not only helps subjects clearly see their learning outcomes and progress, stimulating their learning enthusiasm, but also allows parents, teachers, and other users to more accurately grasp the subject's recent learning status and provide timely supervision and correction.

[0094] In a specific implementation scenario, as the test subjects answer the target questions, the test subjects' level scores and the difficulty scores of the target questions are displayed in real time to encourage them to actively answer the target questions and improve their learning enthusiasm.

[0095] In a specific implementation scenario, the test subjects' ability scores are compared and analyzed with their respective school, region, and other groups to demonstrate their relative position and performance within the group. This relative evaluation not only stimulates the test subjects' competitive spirit but also helps them rationally set learning goals for the next stage, promoting active learning. For teachers and parents, it provides a reference for understanding the test subjects' actual level. It should be noted that the ability scores of each test subject need to be collected with the user's authorization, and based on the test subject's confirmation and choice, their relevant identity information within their school, region, and other groups can be disclosed or not, or corresponding virtual identity information can be set. While disclosing ability scores, personal identity information is not disclosed, and this application does not impose any restrictions on this.

[0096] In one implementation scenario, after each preset period ends, the subject's ability score is updated to maintain the initial level score at the start of a new preset period. Subjects can view their historical semester performance and are encouraged to continue learning, generating motivation for continuous improvement.

[0097] The above scheme uses a grade level to represent the learning level of the test subjects and a difficulty score to represent the difficulty of the questions. When a test subject selects a question as the target question and answers it, the grade level of the test subject and the difficulty score of the target question are obtained. Based on the grade level of the test subject and the difficulty score of the target question, the predicted score rate of the test subject's answer to the target question is obtained, as well as the actual score rate of the test subject's answer to the target question. At least one of the grade level of the test subject and the difficulty score of the target question is updated based on the predicted score rate and the actual score rate. The test subject's answer to each target question will be reflected in the change of the grade level, and the grade level of the test subject is dynamically updated based on the different difficulty levels of each target question, thus improving the accuracy of the learning analysis of the test subjects.

[0098] Please see Figure 6 , Figure 6 This is a schematic diagram of the framework of an embodiment of the learning analysis device 60 of this application. The learning analysis device 60 includes an acquisition module 61 and a generation module 62. The acquisition module 61 is used to acquire the assessment results of the test subjects and the difficulty scores of the target questions; wherein, the assessment results include at least a grade score representing the learning level of the test subjects; the generation module 62 is used to generate the learning analysis results of the test subjects based on the assessment results and the difficulty scores.

[0099] The above scheme generates learning analysis results for the test subjects based on their assessment results and the difficulty scores of the target questions. It uses both the assessment results, which at least represent the test subjects' learning level, and the difficulty scores, which represent the difficulty of the target questions, as reference information for generating the learning analysis results. This allows for learning analysis based on multi-dimensional data, thus improving the accuracy of learning analysis for the test subjects.

[0100] In some disclosed embodiments, the learning analysis results include at least learning comments about the subjects. The generation module 62 also includes a statement selection module (not shown) for selecting target feedback statements and target learning statements based on the assessment results and difficulty scores. The target feedback statements include either encouraging statements or critical statements, and the target learning statements are used to instruct the subjects on the learning content. The generation module 62 also includes a comment generation module (not shown) for generating learning comments based on the target feedback statements and target learning statements.

[0101] In some disclosed embodiments, the evaluation results also include data on the subject's answering of the target question and data on the answering process. The statement selection module (not shown) further includes a candidate selection module (not shown), used to select candidate feedback statements as target feedback statements based on the difficulty score of the target question and the subject's answering result data and grade score; wherein, the candidate feedback statements include several encouraging statements representing different levels of feedback and several critical statements representing different levels of feedback; and, based on the difficulty score of the target question and the subject's answering process data and grade score, to select candidate learning statements as target learning statements; wherein, the candidate learning statements include at least one of answering evaluation statements and learning suggestion statements.

[0102] In some disclosed embodiments, the candidate selection module (not shown) further includes a feedback statement selection module (not shown), used to determine a target baseline value based on whether the subject correctly answered the target question according to the test result data, and to determine a first weight based on the difficulty score and a second weight based on the level score; wherein the first weight is directly proportional to the difficulty score and the second weight is inversely proportional to the level score; the target baseline value is weighted according to the first weight and the second weight respectively to obtain a weighted sum; and a candidate feedback statement with the target feedback degree corresponding to the sum is selected as the target feedback statement.

[0103] In some disclosed embodiments, the acquisition module 61 further includes a grade score acquisition module (not shown), used to acquire the target knowledge points contained in the target questions, and to acquire the subject's historical learning data and historical question-answering data; based on the historical learning data, to determine the subject's historical learning progress regarding the target knowledge points, and based on the historical question-answering data, to determine the subject's historical question-answering accuracy regarding the target knowledge points; and based on the historical learning progress and historical question-answering accuracy, to generate the subject's grade score; wherein, the historical learning progress is directly proportional to the grade score and the historical question-answering accuracy is directly proportional to the grade score.

[0104] In some disclosed embodiments, the acquisition module 61 further includes a difficulty score acquisition module (not shown), used to acquire target test point information contained in the target question; wherein, the target test point information includes at least one of the following: a knowledge graph of the target test point, the difficulty level of the target test point, and the question distribution of the target test point; and based on the target test point information, a difficulty score for the target question is generated.

[0105] In some disclosed embodiments, the generation module 62 further includes a large language generation module (not shown), used to construct prompt text based on the target feedback statement and the target learning statement; wherein, the prompt text is used to indicate the comment generation style of the large language model and to adjust the knowledge text about the learning direction of the subject; send the prompt text to the large language model; and obtain the target statement output by the large language model as the subject's learning comment.

[0106] In some disclosed embodiments, the learning analysis results include at least the updated grade score of the test subject and the updated difficulty score of the target question. The generation module 62 also includes a score rate acquisition module (not shown), which is used to obtain the predicted score rate of the test subject's answer to the target question based on the grade score of the test subject and the difficulty score of the target question, and to obtain the actual score rate of the test subject's answer to the target question. The generation module 62 also includes an update module (not shown), which is used to update at least one of the grade score and the difficulty score based at least on the predicted score rate and the actual score rate.

[0107] In some disclosed embodiments, the level score and difficulty score are obtained based on the Elo rating system. The score rate acquisition module (not shown) also includes a predicted score rate module (not shown), which is used to obtain a first difference between the difficulty score and the level score, and to obtain a target parameter determined by the distribution interval of the Elo rating system for the test subject; wherein, the target parameter characterizes the test subject's answering ability; and the predicted score rate is obtained based on the first difference and the target parameter.

[0108] In some disclosed embodiments, the learning analysis device 60 further includes a first adjustment module (not shown), used to obtain a first adjustment coefficient before updating at least one of the grade score and difficulty score based on at least the predicted score rate and the actual score rate; wherein, the first adjustment coefficient represents the step size of the grade score update; updating at least one of the grade score and difficulty score based on at least the predicted score rate and the actual score rate includes: determining a first adjustment score based on the predicted score rate, the actual score rate and the first adjustment coefficient; and updating the grade score based on the first adjustment score.

[0109] In some disclosed embodiments, the learning analysis device 60 further includes a second adjustment module (not shown) for obtaining a second adjustment coefficient before updating at least one of the grade score and difficulty score based on at least the predicted score rate and the actual score rate; wherein the second adjustment coefficient represents the step size of the difficulty score update; updating at least one of the grade score and difficulty score based on at least the predicted score rate and the actual score rate includes: determining a second adjustment score based on the predicted score rate, the actual score rate and the second adjustment coefficient; and updating the difficulty score based on the second adjustment score.

[0110] In some disclosed embodiments, the learning analysis device 60 further includes a question recommendation module (not shown), which is used to obtain the latest level score of the subject and the latest difficulty score of each question after updating at least one of the level score and difficulty score based on at least the predicted score rate and the actual score rate; determine the range of difficulty score values ​​for recommending questions to the subject based on the latest level score; and determine the recommended questions based on the range of difficulty score values ​​and the latest difficulty score of each question.

[0111] Please see Figure 7 , Figure 7 This is a schematic diagram of an embodiment of the electronic device 70 of this application. The electronic device 70 includes a memory 71 and a processor 72. The memory 71 stores program instructions, and the processor 72 is used to execute the program instructions to implement the steps in any of the above-described learning analysis method embodiments. For details, please refer to the foregoing disclosed embodiments, which will not be repeated here. The electronic device 70 may include, but is not limited to, servers, smartphones, laptops, tablets, self-service machines, etc., and is not limited here.

[0112] Specifically, processor 72 controls itself and memory 71 to implement the steps in any of the above-described learning analysis method embodiments. Processor 72 can also be referred to as a CPU (Central Processing Unit). Processor 72 may be an integrated circuit chip with signal processing capabilities. Processor 72 can also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor. Furthermore, processor 72 can be implemented using integrated circuit chips.

[0113] In the above scheme, the electronic device 70 generates the learning analysis results of the test subjects based on the evaluation results of the test subjects and the difficulty score of the target questions. The evaluation results, which at least represent the learning level of the test subjects, and the difficulty score, which represents the difficulty of the target questions, are used together as reference information for generating the learning analysis results. The learning analysis can be carried out based on multi-angle data, thus improving the accuracy of the learning analysis of the test subjects.

[0114] Please see Figure 8 , Figure 8 This is a schematic diagram of a framework of an embodiment of the computer-readable storage medium 80 of this application. The computer-readable storage medium 80 stores program instructions 81 that can be executed by a processor. The program instructions 81 are used to implement the steps in any of the above-described embodiments of the learning analysis method.

[0115] In the above scheme, the computer-readable storage medium 80 generates the learning analysis results of the test subjects based on the evaluation results of the test subjects and the difficulty scores of the target questions. The evaluation results, which at least represent the learning level of the test subjects, and the difficulty scores, which represent the difficulty of the target questions, are used together as reference information for generating the learning analysis results. This allows for learning analysis based on multi-angle data, thus improving the accuracy of learning analysis of the test subjects.

[0116] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0117] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0119] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0120] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0121] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] If the technical solution of this application involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the individual through pop-up information or by asking the individual to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

Claims

1. A method for analyzing student learning, characterized in that, include: Obtain the assessment results of the test subjects and the difficulty scores of the target questions; wherein, the assessment results include at least a grade score representing the learning level of the test subjects; Based on the subject's rating score and the target question's difficulty score, the predicted score rate of the subject's answer to the target question is obtained, and the actual score rate of the subject's answer to the target question is also obtained. The rating score and the difficulty score are updated based at least on the predicted score rate and the actual score rate; Updating the grade score includes: A first adjustment coefficient is obtained, and the sub-period to which the subject currently answers the target question is determined as the target sub-period; wherein, the first adjustment coefficient is determined based on the total number of questions answered by the subject within a preset period, the preset period includes several sub-periods, and the first adjustment coefficient represents the score update step of the grade score; Based on the total number of questions answered within the target sub-cycle, determine the second cumulative number of questions answered by the subject when currently answering the target question within the target sub-cycle; Based on the grade score and the second cumulative number of questions, a first preset coefficient is determined; Based on the predicted score rate, the actual score rate, the first preset coefficient, and the first adjustment coefficient, a first adjustment score is determined; wherein, the first adjustment score represents the amount of change used to update the grade score, and the first preset coefficient represents the step size of the score adjustment of the first adjustment score; The grade score is updated based on the first adjusted score.

2. The method according to claim 1, characterized in that, The grade score and the difficulty score are obtained based on the Elo grade system. The process of obtaining the predicted score rate of the subject's answer to the target question based on the subject's grade score and the difficulty score of the target question includes: Obtain the first difference between the difficulty score and the level score, and obtain the target parameter determined for the subject based on the distribution interval of the Elo level score system; wherein the target parameter characterizes the subject's answering ability; The prediction score rate is obtained based on the first difference and the target parameter.

3. The method according to claim 1, characterized in that, Obtaining the first adjustment coefficient includes: Based on the total number of questions answered by the subject within a preset period, determine the first cumulative number of questions answered by the subject when answering the target question within the preset period; The first adjustment coefficient is determined based on the first cumulative number of questions.

4. The method according to claim 1, characterized in that, The step of determining the first adjusted score based on the predicted score rate, the actual score rate, and the first adjustment coefficient includes: Obtain the second difference between the actual score rate and the predicted score rate; The first adjusted score is obtained based at least on the product of the second difference and the first adjustment coefficient.

5. The method according to claim 4, characterized in that, The step of obtaining the first adjusted score based at least on the product of the second difference and the first adjustment coefficient includes: Based on the mapping relationship between the subject's grade score and the grade set, the subject's target grade is determined; wherein, the grade set contains several candidate grades; Based on the target level, a second preset coefficient is determined; wherein, the second preset coefficient represents the adjustment step of the first adjustment score; The first adjustment score is determined based on the product of the second difference, the first adjustment coefficient, and the second preset coefficient.

6. The method according to claim 1, characterized in that, Before updating the rating score and the difficulty score based at least on the predicted score rate and the actual score rate, the method further includes: Obtain a second adjustment coefficient; wherein the second adjustment coefficient represents the step size of the difficulty score update; The step of updating the grade score and the difficulty score based at least on the predicted score rate and the actual score rate includes: The second adjusted score is determined based on the predicted score rate, the actual score rate, and the second adjustment coefficient; The difficulty score is updated based on the second adjustment score.

7. The method according to any one of claims 1 to 6, characterized in that, After updating the rating score and the difficulty score based at least on the predicted score rate and the actual score rate, the method further includes: Obtain the latest grade score and the latest difficulty score of each question for the subject. Based on the latest grade scores, the range of difficulty scores for recommending questions to test subjects is determined. Based on the aforementioned difficulty score range and the latest difficulty score for each question, recommended questions are determined.

8. The method according to claim 7, characterized in that, After determining recommended questions based on the difficulty score range and the latest difficulty score of each question, the method further includes: In response to the subject's instruction to select the recommended question, the system obtains the level score update coefficient, selects the recommended question as the new target question, and performs the steps of obtaining the subject's level score and the difficulty score of the target question, as well as subsequent steps, to update the subject's level score and the difficulty score of the recommended question. The updated level score is obtained based on the level score update coefficient, the predicted score rate, and the actual score rate.

9. The method according to any one of claims 1 to 6, characterized in that, After updating the grade score and the difficulty score based at least on the predicted score rate and the actual score rate, the method further includes: Based on the updated level scores obtained by the test subjects after each round of question answering, a trend graph of the test subjects' level scores is generated. The trend graph is shown.

10. A learning analysis device, characterized in that, include: The acquisition module is used to acquire the assessment results of the test subjects and the difficulty scores of the target questions; wherein, the assessment results include at least a grade score representing the learning level of the test subjects; The prediction module is used to obtain the predicted score rate of the subject's answer to the target question based on the subject's level score and the difficulty score of the target question, and to obtain the actual score rate of the subject's answer to the target question. The calculation module is used to update the grade score and the difficulty score based at least on the predicted score rate and the actual score rate; The first adjustment coefficient determination module is used to obtain the first adjustment coefficient and determine the sub-cycle to which the subject currently answers the target question, as the target sub-cycle; wherein, the first adjustment coefficient is determined based on the total number of questions answered by the subject within a preset cycle, the preset cycle includes several sub-cycles, and the first adjustment coefficient represents the score update step of the grade score; The second cumulative question count determination module is used to determine the second cumulative question count of the subject when answering the target question in the target sub-period, based on the total number of questions answered in the target sub-period. The first preset coefficient determination module is used to determine the first preset coefficient based on the grade score and the second cumulative number of questions; The first adjustment score determination module is used to determine a first adjustment score based on the predicted score rate, the actual score rate, the first preset coefficient, and the first adjustment coefficient; wherein, the first adjustment score represents the amount of change used to update the grade score, and the first preset coefficient represents the step size of the score adjustment of the first adjustment score; The grade score update module is used to update the grade score based on the first adjusted score.

11. An electronic device, characterized in that, It includes a memory and a processor that are coupled to each other, the processor being used to execute program instructions stored in the memory to implement the learning analysis method according to any one of claims 1 to 9.

12. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by the processor, they implement the learning analysis method according to any one of claims 1 to 9.

Images