A learning ability evaluation and improvement system based on large models

Abstract

The present invention belongs to the technical field of learning ability assessment, and specifically relates to a learning ability assessment and improvement system based on large models. It includes a data collection module for collecting the historical answering information of users; a calculation module for calculating the historical answering scores of users; through a data processing module, based on the answering accuracy of the same knowledge point, the scoring rate of users for the same knowledge point is calculated, so as to reflect whether there is contingency when users answer questions for the same knowledge point. Based on the calculated scoring rate, the learning ability of users is evaluated through an evaluation module. Secondly, in order to identify the weaknesses of users and provide relevant learning materials or practice questions to focus on strengthening those weak knowledge points, the historical answering proficiency of users is also output through an improvement model. By obtaining the historical answering proficiency of all users, the user groups with relatively weak mastery of knowledge points are screened out, and then the learning of this part of the user groups is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of learning ability assessment, and specifically relates to a learning ability assessment and improvement system based on a large model. Background Art

[0002] Traditional examinations and tests use a single score to evaluate the learning ability of users. It can neither know which specific aspects of knowledge the user has mastered or not mastered, nor can it obtain the reasons why the user got wrong answers to the questions for remedy. For users with the same score, it is even more impossible to obtain the possible differences in their cognitive states and knowledge structures. The information provided by traditional examinations is no longer suitable for the needs of individual development.

[0003] In order to obtain test results that can better reflect the learning ability of users, a patent with the publication number CN114491050A discloses a learning ability assessment method based on cognitive diagnosis, including: obtaining the answering information of users, preprocessing the answering information of users by tagging to obtain tagged answering information and untagged answering information; clustering the untagged answering information according to the tagged answering information to obtain the tags of all answering information; inputting all answering information and their tags into a cognitive diagnosis model, the cognitive diagnosis model outputs the answering correct probability of users, and the learning ability of users is evaluated according to the answering correct probability of users.

[0004] Based on the above, essentially the prior art analyzes the answering status of users during the actual test process by obtaining the answering information of users and using the relevant information in the answering information, so as to comprehensively calculate the true score situation of users, and does not rely on the scores on paper in the traditional sense. However, in the actual calculation process, the prior art can only analyze and obtain the historical true score situation of users based on historical answering information, and simultaneously complete the learning ability assessment of all users. However, during the process of users answering questions, the questions corresponding to the same knowledge point are not unique. Therefore, when users answer multiple questions corresponding to the same knowledge point, if there are both correct and incorrect situations at the same time, there must be a certain contingency in the questions answered by users. If the existence of contingency is not considered, it may lead to the actual true scores of some users being too high and not being able to truly reflect the learning ability of users.

[0005] Therefore, the present invention provides a learning ability assessment and improvement system based on a large model. Summary of the Invention

[0006] In order to make up for the deficiencies of the prior art and solve at least one technical problem proposed in the background art.

[0007] The technical solution adopted by the present invention to solve its technical problems is: A learning ability assessment and improvement system based on a large model according to the present invention includes:

[0008] A data collection module for collecting the user's historical answering information;

[0009] A calculation module for calculating the user's historical answering score based on the historical answering information;

[0010] A data processing module for obtaining the user's historical scoring rate based on the historical answering score and the historical answering information;

[0011] An evaluation module for evaluating the user's historical learning ability based on the historical scoring rate;

[0012] An improvement model for outputting the user's historical answering proficiency based on the input user's historical answering information, and improving the user's learning ability based on the user's historical answering proficiency.

[0013] Preferably, the answering information includes: answering accuracy A, number of answering times C, and answering time T; when the user answers correctly, A = 1, otherwise A = 0.

[0014] Preferably, the method for the calculation module to obtain the historical answering score according to the historical answering information is:

[0015] Obtain the user's answering information;

[0016] According to the formula:

[0017]

[0018] where P represents the basic score, represents the answering accuracy of any question i, represents the number of answering times of any question i; represents the historical answering score of any user j for any question i;

[0019] Calculate the historical answering score of any user j for any question i .

[0020] Preferably, the method for obtaining the historical scoring rate is:

[0021] Obtain the historical answering score and the corresponding historical answering information;

[0022] According to the formula:

[0023]

[0024] where, represents the number of questions answered correctly, n represents the total number of all questions, represents the total theoretical score of all questions; represents the historical scoring rate of any user j;

[0025] The historical scoring rate is calculated .

[0026] Preferably, based on the historical scoring rate , the method for evaluating the user's learning ability is as follows:

[0027] According to the obtained historical scoring rate , the first threshold R1 and the second threshold R2 are set;

[0028] The user's learning ability is evaluated by setting the first threshold R1 and the second threshold R2:

[0029] When > R1, the corresponding user is marked as excellent;

[0030] When < R2, the corresponding user is marked as poor;

[0031] When R2 ≤ ≤ R1, the corresponding user is marked as medium;

[0032] Among them, the first threshold R1 is greater than the second threshold R2.

[0033] Preferably, the calculation module is further configured to calculate the user's historical proficiency in knowledge points;

[0034] It further includes an analysis module, which analyzes the user's mastery of knowledge points based on the user's historical proficiency in knowledge points.

[0035] Preferably, the method for calculating the user's historical proficiency in knowledge points is as follows:

[0036] Obtain the user's historical answering information;

[0037] According to the formula:

[0038]

[0039] Among them, Tstd represents the total standard answering time, represents the answering time of any question, is the user's total answering time, represents the proficiency of any user j; represents the weight corresponding to the answering scoring rate, represents the weight corresponding to the answering time, where > ;

[0040] The historical proficiency of any user j in knowledge points is calculated .

[0041] Preferably, the method for the analysis module to analyze the user's mastery of knowledge points based on historical proficiency is as follows:

[0042] Set the third threshold R3 and the fourth threshold R4;

[0043] When > R3, then mark all the corresponding users as proficient;

[0044] When < R4, then mark all the corresponding users as unfamiliar;

[0045] Among them, the third threshold R3 is greater than the fourth threshold R4;

[0046] When R4 ≤ ≤ R3, then use the K-means algorithm to divide the proficiency between the third threshold and the fourth threshold where the proficiency greater than the third threshold R3 and the proficiency less than or equal to the fourth threshold R4 are used as the initial clusters.

[0047] Preferably, the data collection module is further configured to collect the current answering information of all users marked as unfamiliar; the data processing module is further configured to calculate the current scoring rate of the user according to the current answering information; the improvement model is further configured to calculate the current proficiency of the user based on the current answering information and the current scoring rate.

[0048] Preferably, the calculation module is further configured to calculate the improvement efficiency value , and the method is as follows:

[0049] Obtain the historical proficiency of the user and the current proficiency , the historical scoring rate and the current scoring rate ;

[0050] According to the formula:

[0051]

[0052] where, represents the influence factor of proficiency, represents the influence factor of scoring rate, represents the interval time;

[0053] Calculate the improvement efficiency value of the user per unit time.

[0054] The beneficial effects of the present invention are as follows:

[0055] 1. A learning ability evaluation and improvement system based on a large model according to the present invention calculates the scoring rate of a user for the same knowledge point through a data processing module based on the accuracy of answering questions for the same knowledge point, so as to reflect whether there is contingency when the user answers questions for the same knowledge point. Based on the calculated scoring rate, the evaluation module then evaluates the learning ability of the user. Secondly, in order to identify the user's weaknesses and provide relevant learning materials or practice questions to focus on strengthening those weak knowledge points, the proficiency of the user's historical answers is also output through an improvement model. By obtaining the historical answer proficiency of all users, users with relatively weak mastery of knowledge points are screened out, and then this part of the user group is given improvement learning, thereby forming a closed loop of learning evaluation and improvement;

[0056] 2. A learning ability evaluation and improvement system based on a large model according to the present invention reduces the problem that some users have relatively high historical answer scores caused by contingency through the calculation of the scoring rate . Specifically, by obtaining all the question answering information corresponding to the same knowledge point of the user, that is, obtaining the number of questions answered correctly and the total number of questions n, by introducing , the problem that the scoring rate is relatively high caused by some users having relatively high historical answer scores caused by contingency is reduced; then all users are marked with a first threshold R1 and a second threshold R2, including three grades of excellent, poor and medium, and based on this, the basic learning ability of the user is known. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The present invention will be further described below with reference to the accompanying drawings.

[0058] Figure 1 is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] In order to make the technical means, creative features, achieved purposes and effects of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0060] As Figure 1 shown, a learning ability evaluation and improvement system based on a large model according to an embodiment of the present invention includes:

[0061] A data collection module for collecting the historical answer information of users;

[0062] A calculation module for calculating the historical answer scores of users according to the historical answer information;

[0063] A data processing module for obtaining the historical scoring rate of users based on the historical answer scores and historical answer information;

[0064] An evaluation module that evaluates the user's historical learning ability based on the historical scoring rate;

[0065] An improvement model that outputs the user's historical answering proficiency based on the input user historical answering information, and improves the user's learning ability based on the user historical answering proficiency.

[0066] Based on the above, in essence, the prior art analyzes the answering status of the user during the actual test by obtaining the user's answering information and using the relevant information in the answering information, so as to comprehensively calculate the user's true score, and does not rely on the score on paper in the traditional sense. In the actual calculation process, the prior art can only analyze the user's historical true score based on the historical answering information and simultaneously complete the evaluation of the learning ability of all users. However, during the user's answering process, the questions corresponding to the same knowledge point are not unique. Therefore, when the user answers multiple questions of the same knowledge point, if there are both incorrect and correct situations, there must be a certain degree of contingency in the user's answering of the questions. If the existence of contingency is not considered, it may lead to an overestimation of the actual true score of some users and cannot truly reflect the user's learning ability.

[0067] In an embodiment of the present invention, it further includes a test system for the user to take a test and a database capable of storing a large number of test questions. When the user conducts a stage test, the test system randomly combines N test questions to form a test paper, and any user answers the test paper based on this. The data collection module obtains the user's historical answering information, and the calculation module calculates the user's historical answering score. Then, through the data processing module, based on the answering accuracy of the same knowledge point, the scoring rate of the user for the same knowledge point is calculated to reflect whether there is contingency when the user answers questions of the same knowledge point. Based on the calculated scoring rate, the evaluation module evaluates the user's learning ability. Specifically, the user's learning ability is divided into three levels based on the scoring rate, namely excellent, medium, and poor. Secondly, in order to identify the user's weaknesses and provide relevant learning materials or practice questions to focus on strengthening those weak knowledge points, the improvement model also outputs the user's historical answering proficiency. By obtaining the historical answering proficiency of all users, the user groups with relatively weak mastery of knowledge points are screened out, and then the learning of this part of the user groups is improved to form a closed loop of learning evaluation and improvement;

[0068] It should be noted that when the user enters the test page of the test system, the test paper composed of N test questions randomly combined by the test system will be displayed, and whenever the user enters the test page again, the test paper will be randomly generated again. And when the user leaves the test page, it is default that the user has completed the test, and the user's historical answering score is calculated.

[0069] In one embodiment, the answering information includes: answering accuracy A, number of answering attempts C, and answering time T; when the user answers correctly, A = 1, otherwise A = 0.

[0070] As an embodiment of the present invention, based on obtaining the answering information of any question of the user, including answering accuracy A, number of answering attempts C, and answering time T, it can more truly reflect the real score of the user for any question. Through the expression of answering accuracy A, if the user finally answers incorrectly, it means that the user has not fully mastered the knowledge point. If the answer is correct, then based on the number of answering attempts C and answering time T, analyze the user's mastery of this knowledge point. When the user submits the same question multiple times, that is, the number of answering attempts C > 1, it indicates that the user has uncertainty about this question, and thus reflects that the user may not have fully mastered the knowledge point. At the same time, if the user's answering time T for any question is too long, it reflects that the user may not have fully mastered the knowledge point;

[0071] Among them, the answering information refers to the information of the user answering any question, including answering accuracy A, number of answering attempts C, and answering time T.

[0072] In one embodiment, the method for the calculation module to obtain the historical answering score according to the historical answering information is:

[0073] Obtain the answering information of the user;

[0074] According to the formula:

[0075]

[0076] where P represents the basic score, represents the answering accuracy of any question i, represents the number of answering attempts of any question i; represents the historical answering score of any user j for any question i;

[0077] Calculate the historical answering score of any user j for any question i .

[0078] Based on the above, since the test papers composed of randomly combined N test questions in the database are regenerated every time a user enters the test page, the test questions are not exactly the same each time the user takes the test. And according to the number of test questions stored in the database, the higher the number of test questions, the lower the repetition rate of the test questions regenerated each time. When calculating the historical answering scores of the user each time, the calculation module first preferentially obtains the historical answering scores of the user according to the historical answering information. As an embodiment of the present invention, the answering types include objective questions such as multiple-choice questions, fill-in-the-blank questions, and true-or-false questions. When the user answers, there is no need to input the answering process, that is, if the user answers correctly, they get points, and if they answer wrongly, they lose points. Based on the number of answering times, when the user answers correctly, when the number of answering times C > 1, the user's final historical answering score will show a gradually decreasing trend, that is, after the user submits an answer to any question more than once, the score of that question will necessarily be less than the basic score P. When the user answers incorrectly, regardless of the number of answering times, the historical answering score of that question is 0.

[0079] In one embodiment, the method for obtaining the historical scoring rate is as follows:

[0080] Obtain the historical answering score and the corresponding historical answering information;

[0081] According to the formula:

[0082]

[0083] where, represents the number of questions answered correctly, n represents the total number of all questions, represents the total theoretical score of all questions; represents the historical scoring rate of any user j;

[0084] Calculate to obtain the historical scoring rate .

[0085] Based on the above, since the questions corresponding to the same knowledge point are not unique, and when calculating the historical answering scores of the user, relying only on the basic score P and the number of answering times C cannot accurately calculate the user's true score. Since the same knowledge point may correspond to multiple questions, and if the user fails to master the knowledge point proficiently, there will necessarily be situations where some questions are answered wrongly and some are answered correctly, that is, the above-mentioned contingency. In order to eliminate the problem that the historical answering scores of some users are relatively high caused by contingency, in one embodiment of the present invention, based on the calculation of the scoring rate , reduce the problem that the historical answering scores of some users are relatively high caused by contingency. Specifically, by obtaining all the answering information of the questions corresponding to the same knowledge point of the user, that is, obtaining the number of questions answered correctly and the total number of all questions n, by introducing , reducing the problem that the high historical answering scores of some users caused by chance lead to a high scoring rate.

[0086] In one embodiment, based on the historical scoring rate , the method for evaluating the user's learning ability is as follows:

[0087] According to the obtained historical scoring rate , set the first threshold R1 and the second threshold R2;

[0088] Evaluate the user's learning ability by setting the first threshold R1 and the second threshold R2:

[0089] When > R1, then mark the corresponding user as excellent;

[0090] When < R2, then mark the corresponding user as poor;

[0091] When R2 ≤ ≤ R1, then mark the corresponding user as medium;

[0092] Among them, the first threshold R1 is greater than the second threshold R2.

[0093] Based on the above, according to the obtained historical scoring rate of the user , and then evaluate the user's learning ability. In one embodiment of the present invention, the first threshold R1 and the second threshold R2 are set, and all users are marked through the first threshold R1 and the second threshold R2, including three grades: excellent, poor, and medium. Based on this, the user's basic learning ability can be known.

[0094] In one embodiment, the calculation module is further used to calculate the user's historical proficiency in knowledge points;

[0095] It further includes an analysis module, which analyzes the user's mastery of knowledge points based on the user's historical proficiency in knowledge points.

[0096] In order to obtain the user's mastery of knowledge points, it is also necessary to calculate the proficiency. The calculation module is used to calculate the user's proficiency in knowledge points, and then analyze the user's mastery of knowledge points through the historical proficiency. Different from the learning ability, the above scoring rate is used to represent the user's basic learning ability and initially reflect the user's mastery of knowledge points, but it still cannot accurately represent the proficiency of all users in knowledge points. For example, in the case of a high user scoring rate, if one user has a high answering accuracy but a large number of submissions, while another user has a low answering accuracy but a small number of submissions; for the above two users, only through the scoring rate, it is impossible to distinguish their mastery of knowledge points.

[0097] In one embodiment, the method for calculating the user's historical proficiency in knowledge points is as follows:

[0098] Obtain the user's historical answering information;

[0099] According to the formula:

[0100]

[0101] where Tstd represents the total standard answering time, represents the answering time of any question, is the user's total answering time, represents the proficiency of any user j; represents the weight corresponding to the answering score rate, represents the weight corresponding to the answering time, where, > ;

[0102] Calculate the historical proficiency of any user j in the knowledge point .

[0103] Based on the above, in order to represent the user's mastery of knowledge points through data, in an embodiment of the present invention, the historical proficiency is obtained through a formula , and the data is used to represent the mastery of any user in the knowledge point. Based on the above formula, by setting the ratio of the user's total answering time to the total standard answering time Tstd, when this ratio is larger, the user's historical proficiency is relatively smaller, and vice versa. At the same time, in order to eliminate the relative influence of the score rate and the answering time, that is, when there is any user with a high score rate and a long answering time or a slightly lower score rate but a short answering time, the historical proficiency calculated in both cases may have a certain overlap. In order to avoid the above situation, different weights are assigned to the score rate and the answering time. In an embodiment of the present invention, it is set that > , indicating that the influence of the score rate is greater than the influence of the answering time.

[0104] In one embodiment, the method for the analysis module to analyze the user's mastery of knowledge points based on historical proficiency is as follows:

[0105] Set the third threshold R3 and the fourth threshold R4;

[0106] When >R3, then all the corresponding users are marked as proficient;

[0107] When < R4, then all the corresponding users are marked as unfamiliar;

[0108] where the third threshold R3 is greater than the fourth threshold R4;

[0109] When R4 ≤ ≤ R3, the K-means algorithm is used to divide the proficiency levels between the third threshold and the fourth threshold where the proficiency levels greater than the third threshold R3 and the proficiency levels less than or equal to the fourth threshold R4 are used as the initial clusters.

[0110] As an embodiment of the present invention, the third threshold R3 and the fourth threshold R4 are set to classify the calculated historical proficiency levels corresponding to all users so as to know which users need to be improved through learning. And there are some users' historical proficiency levels existing in the interval [R4, R3]. Therefore, the K-means algorithm still needs to be used to divide the proficiency levels between the third threshold and the fourth threshold so that all users can finally be marked as proficient or unfamiliar, and then the unfamiliar users are to be improved through learning, thus completing the closed-loop from evaluation to improvement.

[0111] In one embodiment, the data collection module is further configured to collect the current answering information of all users marked as unfamiliar; the data processing module is further configured to calculate the current scoring rate of the users according to the current answering information; and the improvement model is further configured to calculate the current proficiency level of the users based on the current answering information and the current scoring rate.

[0112] As an embodiment of the present invention, the calculated historical answering scores, historical scoring rates, and historical proficiency levels of the users are all used as historical records and can be stored in the data storage unit. In order to verify whether the users have mastered the knowledge points proficiently after being improved through learning, further tests are needed; among them, the data collection module further includes a data storage unit.

[0113] In one embodiment, the calculation module is further configured to calculate the improvement efficiency value , and the method is as follows:

[0114] Obtain the historical proficiency level and the current proficiency level of the user, as well as the historical scoring rate and the current scoring rate ;

[0115] According to the formula:

[0116]

[0117] Among them, represents the influence factor of proficiency, represents the influence factor of scoring rate, represents the interval time;

[0118] Calculate the improvement efficiency value of the user per unit time .

[0119] By according to the user's historical proficiency and the current proficiency , historical scoring rate and the current scoring rate Calculate the improvement efficiency value of the user , based on the improvement efficiency value reflects the change in the user's mastery of knowledge points after learning and improvement, which helps the user master the knowledge points more quickly and helps adjust the specific operations of learning and improvement.

[0120] Working principle: The historical answering information of the user is obtained through the data collection module, and the historical answering scores of the user are calculated depending on the calculation module. Then, through the data processing module, based on the answering accuracy of the same knowledge point, the scoring rate of the user for the same knowledge point is calculated, so as to reflect whether there is contingency when the user answers questions for the same knowledge point. Based on the calculated scoring rate, the learning ability of the user is evaluated through the evaluation module. Specifically, the learning ability of the user is divided into three levels based on the scoring rate, namely excellent, medium and poor. Secondly, in order to identify the user's weaknesses, provide relevant learning materials or exercises to focus on strengthening those weak knowledge points. The historical answering proficiency of the user is also output through the improvement model. By obtaining the historical answering proficiency of all users, the user groups with relatively weak mastery of knowledge points are screened out, and then the learning of this part of the user groups is improved, so as to form a closed loop of learning evaluation and improvement; when calculating the historical answering score of the user each time, first the calculation module preferentially obtains the historical answering score of the user according to the historical answering information. As an implementation manner of the present invention, the answering types include objective questions such as multiple-choice questions, fill-in-the-blank questions, and true-or-false questions. When the user answers, there is no need to input the answering process, that is, if the user answers correctly, they get points, and if they answer incorrectly, they lose points. Based on the number of answering times, when the user answers correctly, when the number of answering times C>1, the final historical answering score of the user will show a gradually decreasing trend, that is, after the user submits a question more than once, the score of this question will necessarily be less than the basic score P. When the user answers incorrectly, regardless of the number of answering times, the historical answering score of this question is 0;

[0121] Based on the scoring rate Calculation is performed to reduce the problem that the historical answering scores of some users are relatively high due to contingency. Specifically, by obtaining all the answering information of the questions corresponding to the same knowledge point of the user, that is, obtaining the number of questions answered correctly and the total number of questions n, by introducing , the problem that the scoring rate is relatively high caused by the relatively high historical answering scores of some users due to contingency is reduced; the calculation module is used to calculate the proficiency of the user in the knowledge point, and then the historical proficiency is used to analyze the user's mastery of the knowledge point. Different from the learning ability, the above scoring rate is used to represent the user's basic learning ability and initially reflect the user's mastery of the knowledge point, but it still cannot accurately represent the proficiency of all users in the knowledge point. For example, when the user's scoring rate is high, if one user has a high answering accuracy but a large number of submission times, while another user has a low answering accuracy but a small number of submission times; for the above two users, the scoring rate alone cannot distinguish their mastery of the knowledge point.

[0122] The historical proficiency is obtained through a formula , and data is used to represent the mastery of any user in the knowledge point. Based on the above formula, by setting the ratio of the total answering time of the user to the total standard answering time Tstd, when this ratio is larger, the historical proficiency of the user is relatively smaller, and vice versa. At the same time, in order to eliminate the relative influence of the scoring rate and the answering time, that is, when there is any user with a high scoring rate and a long answering time or a slightly lower scoring rate but a short answering time, the historical proficiency calculated in both cases may have a certain overlap. In order to avoid the above situation, different weights are assigned to the scoring rate and the answering time.

[0123] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A learning ability evaluation and improvement system based on a large model, characterized in that, It includes: A data collection module for collecting the user's historical answering information; A calculation module for calculating the user's historical answering score based on the historical answering information; A data processing module for obtaining the user's historical scoring rate based on the historical answering score and the historical answering information; An evaluation module for evaluating the user's historical learning ability based on the historical scoring rate; An improvement model for outputting the user's historical answering proficiency based on the input user's historical answering information, and improving the user's learning ability according to the user's historical answering proficiency; The answering information includes: answering accuracy A, number of answering times C, and answering time T; when the user answers correctly, A = 1, otherwise A = 0; The method by which the calculation module obtains the historical answering score according to the historical answering information is: Obtain the user's answering information; According to the formula: M ij = P × A i × (1 - 0.1 × (C i - 1)), Among them, P represents the basic score, and A i represents the answering accuracy of any question i, and C i represents the number of times of answering any question i; M ij represents the historical answering score of any question i of any user j; Calculate the historical answering score M of any question i for any user j ij ; The method for obtaining the historical scoring rate is: Obtain the historical answering score and the corresponding historical answering information; According to the formula: where n x represents the number of questions answered correctly, n represents the total number of all questions, represents the total theoretical score of all questions; S j represents the historical scoring rate of any user j; Calculate the historical scoring rate S j ; Based on the calculation of the scoring rate S j The problem that some users have relatively high historical answering scores caused by chance is reduced. Specifically, by obtaining all the answering information of the questions corresponding to the same knowledge point of the user, that is, obtaining the number of questions n with correct answers x And the total number of questions n, by introducing The problem that the scoring rate is relatively high caused by the relatively high historical answering scores of some users caused by chance is reduced; The calculation module is also used to calculate the user's historical proficiency for knowledge points; It further includes an analysis module for analyzing the user's mastery of knowledge points based on the user's historical proficiency for knowledge points; The method for calculating the user's historical proficiency for knowledge points is: Obtain the user's historical answering information; According to the formula: Among them, Tstd represents the total standard answering time, and T i represents the answering time for any question, is the total answering time of the user, U j represents the proficiency of any user j; α represents the weight corresponding to the answering score rate, and 1 - α represents the weight corresponding to the answering time, where α > 1 - α; Calculate the historical proficiency U of any user j for the knowledge point j ; Obtain the historical proficiency U through a formula j , use data to represent the mastery degree of any user for knowledge points. Based on the above formula, by setting the total answering time of the user and the ratio of the total answering standard time Tstd, when this ratio is larger, the historical proficiency U of the user j is relatively smaller, and vice versa. At the same time, in order to eliminate the relative influence of the scoring rate and the answering time, that is, when there is any user with a high scoring rate and a long answering time or a slightly lower scoring rate but a short answering time, the historical proficiency U j calculated in both cases may have a certain overlap. In order to avoid the above situation, different weights are assigned to the scoring rate S j and the answering time.

2. The learning ability evaluation and improvement system based on a large model according to claim 1, wherein: Based on the historical scoring rate S j , the method for evaluating the user's learning ability is as follows: According to the obtained historical scoring rate S j , set a first threshold R1 and a second threshold R2; The user's learning ability is evaluated by setting a first threshold R1 and a second threshold R2: When S j > R1, the corresponding user is marked as excellent; When S j < R2, the corresponding user is marked as poor; When R2 ≤ S j ≤ R1, the corresponding user is marked as medium; Among them, the first threshold R1 is greater than the second threshold R2.

3. The learning ability evaluation and improvement system based on a large model according to claim 1, characterized in that: The method by which the analysis module analyzes the user's mastery of knowledge points based on the historical proficiency is: Set a third threshold R3 and a fourth threshold R4; When U j > R3, then all the corresponding users are marked as proficient; When U j < R4, then all the corresponding users are marked as unfamiliar; Among them, the third threshold R3 is greater than the fourth threshold R4; When R4 ≤ U j ≤ R3, the K-means algorithm is used to partition the proficiency U between the third threshold and the fourth threshold j , where the proficiency U greater than the third threshold R3 j and the proficiency U less than or equal to the fourth threshold R4 j are used as the initial clusters.

4. The learning ability evaluation and improvement system based on a large model according to claim 3, characterized in that: The data collection module is also used to collect the current answering information of all users marked as unfamiliar; the data processing module is also used to calculate the user's current scoring rate according to the current answering information; the improvement model is also used to calculate the user's current proficiency based on the current answering information and the current scoring rate.

5. The learning ability evaluation and improvement system based on a large model according to claim 4, characterized in that: The calculation module is also used to calculate the improvement efficiency value G j , and the method is as follows: Obtain the user's historical proficiency U j and the current proficiency U j * , historical scoring rate S j and the current scoring rate S j * ; According to the formula: Among them, δ represents the influence factor of proficiency, θ represents the influence factor of the scoring rate, and ΔT represents the interval time; Calculate the efficiency improvement value G of the user per unit time j .

Images