An intelligent question search and adaptive recommendation system

Abstract

This invention relates to the field of adaptive question recommendation technology, and more particularly to an intelligent question search and adaptive recommendation system. This system recommends relevant courses to users by statistically analyzing question search information. It can make more accurate recommendations based on user behavior, and adjusts the proportion of test questions pushed to a single subject according to the number of knowledge points in each subject. After adjusting the proportion of test questions pushed to a single subject, it adaptively adjusts the proportion of test questions pushed to other subjects, thereby recommending test questions more effectively. Simultaneously, it determines the suitability of the recommended test questions based on the user's answers to the recommended test questions, and generates corresponding adjustment instructions based on the reasons for inadequacy. This allows for more accurate determination of the effectiveness of adaptively recommended test questions and more effective adjustments based on the reasons for inadequacy, thereby further improving the user's learning efficiency.

Description

Technical Field

[0001] This invention relates to the field of adaptive question recommendation technology, and in particular to an intelligent question search and adaptive recommendation system. Background Technology

[0002] Intelligent question search and adaptive recommendation systems can "precisely match needs" and "dynamically adapt to changes," saving time and improving learning efficiency. They can also optimize resource allocation and improve educational equity. At the same time, they can promote the deep integration of AI and education, providing infrastructure for future education models (such as lifelong learning and skills reshaping).

[0003] Chinese Patent Publication No. CN109063182B discloses a content recommendation method and electronic device based on voice-based question search. The method includes: extracting a target question from the input voice; determining whether an answer matching the target question has been found; if no answer matching the target question has been found, determining the target subject to which the target question belongs based on keywords identified from the target question; determining the target content with the highest matching degree to the target question from all teaching content corresponding to the target subject; and outputting the target content.

[0004] It is evident that the existing technology has the following problems: it fails to determine whether the test questions are qualified based on the user's answers to the recommended test questions, and it also fails to adjust the test questions based on the reasons for their failure. As a result, the effectiveness of the adaptively recommended test questions cannot be determined, leading to low learning efficiency for users. Summary of the Invention

[0005] To address this, the present invention provides an intelligent question search and adaptive recommendation system to overcome the problems in the prior art where the system fails to determine whether a test question is qualified based on the user's answer information to the recommended test questions, and also fails to adjust the system based on the reasons for the unqualified answer. As a result, the system cannot determine the effectiveness of the adaptively recommended test questions, which reduces the efficiency of the push notification and leads to lower learning efficiency for the user.

[0006] To achieve the above objectives, the present invention provides an intelligent question search and adaptive recommendation system, comprising:

[0007] The recognition module is used to recognize the questions entered by the user;

[0008] An output module, connected to the recognition module, is used to acquire and output the answer corresponding to the question through big data.

[0009] The statistics module, which is connected to the output module, is used to record and statistically analyze the search information during the search process, including the subject and knowledge points.

[0010] The push module, which is connected to the statistics module, is used to push relevant test questions to the user based on the search information;

[0011] The recording module, which is connected to the push module, is used to record the user's answer information when doing test questions, including the answering time and the score rate;

[0012] An analysis module, connected to the recording module, is used to determine whether the push of test questions is qualified based on the obtained answer information, and to generate corresponding adjustment instructions based on the reasons for non-compliance, wherein the adjustment instructions include adjusting the percentage of qualified subjects and adjusting the total number of pushed questions.

[0013] An adjustment module, connected to the analysis module, is used to adjust the corresponding parameter to the corresponding value based on the received adjustment command.

[0014] Furthermore, the push module is also used to mark the proportion of each subject type searched within a preset time period as the proportion of the number of pushes for each subject during the push process; the push module is also used to adjust the proportion of the number of pushes for each subject according to the number of knowledge point types in each subject; the push module is also used to increase the proportion of the number of pushes for each subject based on the number of knowledge point types, and the increase in the proportion of the number of pushes is proportional to the increase in the number of knowledge point types.

[0015] Furthermore, the push module is also used to adjust the push quantity ratio of each subject sequentially based on the adjusted push quantity ratio of a single subject to ensure that the sum of the push quantity ratios of each subject is 1; the push module is also used to determine the adjustment order of the push quantity ratio of each subject based on the ascending order of the number of knowledge point types contained in all search questions of each subject; the push module also sets a critical push quantity ratio, and the adjusted push quantity ratio of each subject is greater than or equal to the critical push quantity ratio.

[0016] Furthermore, the analysis module is also used to determine whether the push notification for the test questions is qualified based on the accuracy rate of a single subject, and to determine the reasons for the unqualified push notification for the test questions based on the user's average answering time or the variance of the accuracy rate of a single knowledge point, wherein the accuracy rate of a single knowledge point is the answering accuracy rate of a user for all questions containing the same knowledge point under that subject.

[0017] Furthermore, the analysis module is also used to generate corresponding processing methods based on the comparison results between the user's average answering time and the preset answering time, including issuing a notification to adjust the proportion of push notifications, or adjusting the total number of push questions based on the difference between the average answering time and the preset answering time.

[0018] Furthermore, the analysis module is also used to reduce the total number of questions pushed based on the difference between the average answering time and the preset answering time, and the difference is inversely proportional to the reduction in the total number of questions pushed.

[0019] Furthermore, the analysis module is also used to generate corresponding processing methods based on the comparison results of the variance of the accuracy of a single knowledge point with the preset variance, including determining the reason for the failure of the pushed questions based on the proportion of qualified subjects, or adjusting the number of related questions for each knowledge point based on the accuracy of each knowledge point, wherein qualified subjects are those whose variance of the accuracy of a single knowledge point is less than or equal to the preset variance.

[0020] Furthermore, the analysis module is also used to generate corresponding processing methods based on the comparison results between the percentage of qualified subjects and the preset percentage, including adjusting the number of qualified subjects pushed based on the absolute value of the slope of the forgetting curve predicted by the LSTM network, or adjusting the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage.

[0021] Furthermore, the analysis module also uses the absolute value of the slope of the forgetting curve predicted by the LSTM network to reduce the number of qualified subjects pushed, and the absolute value of the slope is inversely proportional to the reduction in the number of qualified subjects pushed.

[0022] Furthermore, the analysis module is also used to reduce the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage, and the difference is proportional to the reduction in the total number of questions pushed.

[0023] Compared with existing technologies, the beneficial effects of this invention are as follows: This system recommends relevant courses to users by statistically analyzing the search information during question searching. It can make more accurate recommendations based on user behavior, and adjust the proportion of test questions pushed to a single subject according to the number of knowledge points in each subject. After adjusting the proportion of test questions pushed to a single subject, it adaptively adjusts the proportion of test questions pushed to other subjects, thereby recommending test questions more effectively. At the same time, it determines whether the test question recommendation is qualified based on the user's answer information to the recommended test questions, and generates corresponding adjustment instructions based on the reasons for unqualified answers. This can more accurately determine the effectiveness of adaptively recommended test questions and make more effective adjustments based on the reasons for unqualified answers, thereby further improving the user's learning efficiency.

[0024] Furthermore, the present invention adjusts the proportion of push notifications for each subject based on the difference between the number of knowledge point types in each subject and the preset number. This allows for more accurate adjustment of the proportion of push notifications for each subject, thereby providing users with more precise test recommendations and further improving their learning efficiency.

[0025] Furthermore, the present invention also adjusts the proportion of push notifications for each subject sequentially based on the adjusted proportion of push notifications for a single subject, and determines the adjustment order of the proportion of push notifications for each subject in ascending order of the number of knowledge point types, which enables more accurate adjustment of the proportion of push notifications for other subjects after adjusting the proportion of push notifications for a single subject.

[0026] Furthermore, the present invention also determines whether the test questions are qualified based on the accuracy of a single subject, which can more quickly determine whether the test questions are qualified, thereby more effectively judging whether the push is effective, and further improving the user's learning efficiency.

[0027] Furthermore, the present invention also performs a secondary judgment on whether the pushed test questions are qualified based on the comparison between the user's average answering time and the preset answering time, which can more accurately determine whether the pushed test questions are qualified, thereby further improving the user's learning efficiency.

[0028] Furthermore, the present invention reduces the total number of questions pushed based on the difference between the user's average answering time and the preset answering time, which can more accurately adjust the total number of questions pushed, thereby further improving the efficiency of the push and also further improving the user's learning efficiency.

[0029] Furthermore, based on the comparison between the variance of the accuracy of a single knowledge point and a preset variance, this invention can more accurately determine the reasons for unqualified push questions according to the mastery of each knowledge point, thereby further improving the efficiency of push questions and thus further improving the user's learning efficiency.

[0030] Furthermore, the present invention also determines the reasons for unqualified push questions based on the comparison results of the percentage of qualified subjects with the preset percentage, thereby determining the reasons for unqualified push questions according to the user's mastery of each subject, and subsequently adjusting the push questions more accurately according to the reasons, thereby further improving the user's learning efficiency.

[0031] Furthermore, the present invention adjusts the number of qualified subjects pushed based on the absolute value of the slope of the forgetting curve predicted by the LSTM network. This allows for a gradual reduction in the number of qualified subjects pushed according to the user's memory status, thereby enabling more accurate adjustment of the number of qualified subjects pushed and further improving the user's learning efficiency.

[0032] Furthermore, the present invention adjusts the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage, which can more accurately adjust the total number of questions pushed, thereby making recommendations more effectively based on the user's situation and further improving the user's learning efficiency. Attached Figure Description

[0033] Figure 1This is a schematic diagram of the intelligent question search and adaptive recommendation system according to an embodiment of the present invention;

[0034] Figure 2 This is a flowchart illustrating the steps involved in implementing the intelligent question search and adaptive recommendation system according to an embodiment of the present invention.

[0035] Figure 3 This is a flowchart illustrating the steps for determining the accuracy of a single subject based on a comparison with a preset accuracy rate, according to an embodiment of the present invention.

[0036] Figure 4 This is a flowchart illustrating the steps for determining the comparison result between the variance of the accuracy of a single knowledge point and a preset variance in an embodiment of the present invention. Detailed Implementation

[0037] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0038] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0039] It should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0040] Please see Figure 1 As shown, it is a schematic diagram of the intelligent question search and adaptive recommendation system according to an embodiment of the present invention.

[0041] The system includes an identification module, an output module, a statistics module, a push module, a recording module, an analysis module, and an adjustment module.

[0042] The recognition module is used to recognize the questions entered by the user;

[0043] The output module is connected to the recognition module and is used to acquire and output the answer corresponding to the question through big data.

[0044] The statistics module is connected to the output module and is used to record and statistically analyze the search information during the search process. The search information includes the subject and knowledge points.

[0045] The push module is connected to the statistics module, and it is used to push relevant test questions to users based on the search information.

[0046] The recording module is connected to the push module and is used to record the user's answer information when doing test questions, including the answering time and the score rate.

[0047] The analysis module is connected to the recording module. It is used to determine whether the push of test questions is qualified based on the obtained answer information, and to generate corresponding adjustment instructions based on the reasons for the failure. The adjustment instructions include adjusting the percentage of qualified subjects and adjusting the total number of pushed questions.

[0048] The adjustment module is connected to the analysis module and is used to adjust the corresponding parameter to the corresponding value based on the received adjustment command.

[0049] Specifically, in this embodiment, all recommended questions are multiple-choice. In the recognition module, the user can input questions via photo, text, or voice. The output module then outputs the corresponding answers based on the user's input. Simultaneously, the statistics module compiles and records data based on search records and the corresponding subjects and knowledge points. Furthermore, the push module sends relevant test questions to the user, allowing them to reinforce their knowledge through repeated practice. Next, the recording module records the user's answer information after the push module delivers the questions, such as answering time and score. The analysis module determines whether the recommended test questions are suitable based on the user's answer information, and if not, identifies the reasons and generates corresponding adjustment instructions. Finally, the adjustment module adjusts the corresponding parameters based on the adjustment instructions determined by the reasons.

[0050] Please see Figure 2 The diagram shown is a flowchart illustrating the steps involved in implementing the intelligent question search and adaptive recommendation system according to an embodiment of the present invention.

[0051] The implementation process of an intelligent question search and adaptive recommendation system includes:

[0052] S1, the recognition module identifies the question entered by the user;

[0053] S2, the output module connected to the recognition module acquires and outputs the answer corresponding to the question through big data;

[0054] S3, The statistical module connected to the output module records and statistically analyzes the question search information during the question search process, wherein the question search information includes the subject and knowledge points;

[0055] S4, the push module connected to the statistics module pushes relevant test questions to the user based on the search information;

[0056] S5, the user's answer information when doing test questions is recorded by the recording module connected to the push module, wherein the answer information includes the answering time and the score rate;

[0057] S6, the analysis module connected to the recording module determines whether the push of test questions is qualified based on the obtained answer information, and generates a corresponding adjustment instruction based on the reason for the failure, wherein the adjustment instruction includes adjusting the percentage of qualified subjects and adjusting the total number of pushed questions.

[0058] S7, the adjustment module connected to the analysis module adjusts the corresponding parameter to the corresponding value based on the received adjustment instruction.

[0059] Specifically, in this embodiment of the invention, the push module is further used to mark the proportion of each subject type searched within a preset time period as the proportion of the number of pushes for each subject during the push process; the push module is further used to adjust the proportion of the number of pushes for each subject according to the number of knowledge point types in each subject; the push module is further used to increase the proportion of the number of pushes for each subject based on the difference between the number of knowledge point types and the preset number, and the difference is proportional to the increase in the proportion of the number of pushes.

[0060] Specifically, in this embodiment, the proportion of subjects whose number of knowledge point types exceeds a preset number is adjusted, where the preset difference G0 = 2. The comparison process between the difference G between the number of knowledge point types and the preset number and the preset difference G0 is as follows:

[0061] If the difference G is less than or equal to the preset difference G0, the proportion of each subject's push notification quantity will be adjusted to 1.2 times the original push notification quantity proportion. It should be noted that the adjusted push notification quantity proportion will be rounded up to the nearest integer.

[0062] If the difference G is less than or equal to the preset difference G0, the proportion of each subject's push notification quantity will be adjusted to 1.4 times the original push notification quantity proportion. It should be noted that the adjusted push notification quantity proportion will be rounded up to the nearest integer.

[0063] Specifically, the push module in this embodiment of the invention is further used to adjust the push quantity ratio of each subject sequentially based on the adjusted push quantity ratio of a single subject to ensure that the sum of the push quantity ratios of each subject is 1; the push module is further used to determine the adjustment order of the push quantity ratio of each subject based on the ascending order of the number of knowledge point types contained in all search questions of each subject; the push module also has a critical push quantity ratio, and the adjusted push quantity ratio of each subject is greater than or equal to the critical push quantity ratio.

[0064] Specifically, in this embodiment, when adjusting the proportion of push notifications for a single subject, if the proportion of push notifications for the adjusted subject is less than the critical proportion of push notifications, then the proportion of push notifications for that subject is adjusted to the critical proportion of push notifications, and the proportion of push notifications for the next subject is adjusted according to a determined adjustment order to ensure that the sum of the proportions of push notifications for all subjects is 1.

[0065] Please see Figure 3 The diagram shows a flowchart illustrating the steps of determining the accuracy rate of a single subject based on a preset accuracy rate according to an embodiment of the present invention. The analysis module in this embodiment is further used to determine whether the push notification for test questions is qualified based on the accuracy rate of a single subject, and to determine the reasons for unqualified push notifications based on the user's average answering time or the variance of the accuracy rate of a single knowledge point. The accuracy rate of a single knowledge point is the user's answering accuracy rate for all questions containing the same knowledge point under that subject.

[0066] Specifically, in this embodiment, all the pushed questions are multiple choice questions. The accuracy rate of a single subject can be divided into a first preset accuracy rate L1 and a second preset accuracy rate L2. The accuracy rate standards are set as follows: first preset accuracy rate L1 = 60%, second preset accuracy rate L2 = 85%. It should be noted that in other embodiments, the values ​​of L1 and L2 can also be determined based on the needs of adaptive recommendation. The comparison process between accuracy rate L and L1 and L2 is as follows:

[0067] If the accuracy rate L is less than or equal to the first preset accuracy rate L1, it indicates that the user has a low level of mastery of the subject, and the push of test questions is deemed qualified.

[0068] If the accuracy rate L is greater than the first preset accuracy rate L1 and less than the second preset accuracy rate L2, it means that it is impossible to determine whether there are other factors that cause this result. Then, the reason for the failure of the push for the test questions is determined based on the user's average answering time P.

[0069] If the accuracy rate L is greater than or equal to the second preset accuracy rate L2, it indicates that the user has a good grasp of the subject. If the push notification for the test questions is deemed unqualified, the reason for the unqualified push notification for the test questions is determined based on the variance Q of the accuracy rate of a single knowledge point.

[0070] Specifically, the analysis module described in this embodiment of the invention is also used to generate a corresponding processing method based on the comparison result between the user's average answering time and the preset answering time, including issuing a notification to adjust the proportion of push notifications, or adjusting the total number of push questions based on the difference between the average answering time and the preset answering time.

[0071] Specifically, in this embodiment, the preset answering time P0 = 1 minute, and the comparison process between the average answering time P and the preset answering time P0 is as follows:

[0072] If the average answering time P is greater than the preset answering time P0, it indicates that the user has not fully mastered the subject, and a notification to adjust the proportion of push notifications will be issued.

[0073] If the average answering time P is less than or equal to the preset answering time P0, it means that the user has mastered the subject. Then, the total number of questions pushed is adjusted based on the difference R between the average answering time and the preset answering time.

[0074] Specifically, the analysis module described in this embodiment of the invention is further used to reduce the total number of questions pushed based on the difference between the average answering time and the preset answering time, and the difference is inversely proportional to the reduction in the total number of questions pushed.

[0075] Specifically, in this embodiment, the preset difference R0 between the average answering time and the preset answering time is 10s. The comparison process based on the difference R and the preset difference R0 is as follows:

[0076] If the difference R is less than or equal to the preset difference R0, the total number of push questions will be adjusted to 0.6 times the original total number of push questions. It should be noted that the adjusted total number of push questions will be rounded up.

[0077] If the difference R is greater than the preset difference R0, the total number of pushed questions will be adjusted to 0.9 times the original total number of pushed questions. It should be noted that the adjusted total number of pushed questions will be rounded up.

[0078] Please see Figure 4The diagram shows the steps for determining the accuracy variance of a single knowledge point based on a comparison with a preset variance in an embodiment of the present invention. The analysis module in this embodiment is also used to generate corresponding processing methods based on the comparison results of the accuracy variance of a single knowledge point with the preset variance. These methods include determining the reasons for unqualified questions based on the percentage of qualified subjects, or adjusting the number of related questions for each knowledge point based on its accuracy. Qualified subjects are those whose accuracy variance for a single knowledge point is less than or equal to the preset variance.

[0079] Specifically, in this embodiment, the preset variance Q0 based on the accuracy of a single knowledge point is 0.95. The comparison process between the variance Q based on the accuracy of a single knowledge point and the preset variance Q0 is as follows:

[0080] If the variance Q of the accuracy rate of a single knowledge point is less than or equal to the preset variance Q0, it means that the user has mastered all the knowledge points of the subject. The subject is then marked as a qualified subject, and the reason for the unqualified push questions is determined based on the percentage T of qualified subjects.

[0081] If the variance Q of the accuracy rate of a single knowledge point is greater than the preset variance Q0, it indicates that the user has not mastered any knowledge points in the pushed subjects. In this case, the number of related questions for each knowledge point will be adjusted based on the accuracy U of each knowledge point.

[0082] Specifically, in this embodiment, the process of adjusting the number of related questions for each knowledge point based on its accuracy is as follows: While ensuring that the subject proportion and the total number of questions pushed remain unchanged, a preset accuracy rate U0 = 90% is set for each knowledge point. The comparison process between the accuracy rate U0 of each knowledge point and the preset accuracy rate U0 is as follows:

[0083] If the accuracy rate U is greater than or equal to the preset accuracy rate U0, then the push questions related to that knowledge point are deleted.

[0084] If the accuracy rate U is less than the preset accuracy rate U0, then the number of push questions related to that knowledge point will be increased.

[0085] Specifically, the analysis module described in this embodiment of the invention is also used to generate a corresponding processing method based on the comparison result between the percentage of qualified subjects and the preset percentage, including adjusting the number of qualified subjects pushed based on the absolute value of the slope of the forgetting curve predicted by the LSTM network, or adjusting the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage.

[0086] Specifically, in this embodiment, the preset percentage of qualified subjects T0 = 0.8. The comparison process between the percentage of qualified subjects T and the preset percentage T0 is as follows:

[0087] If the percentage of qualified subjects T is less than or equal to the preset percentage T0, the number of qualified subjects pushed is adjusted based on the absolute value H of the slope of the forgetting curve predicted by the LSTM network.

[0088] If the percentage of qualified subjects T is greater than the preset percentage T0, it indicates that the user has a basic grasp of the pushed questions. Then, the total number of pushed questions is adjusted based on the difference I between the percentage of qualified subjects and the preset percentage.

[0089] Specifically, the analysis module described in this embodiment of the invention is also used to reduce the number of qualified subjects pushed based on the absolute value of the slope of the forgetting curve predicted by the LSTM network, and the absolute value of the slope is inversely proportional to the reduction in the number of qualified subjects pushed.

[0090] Specifically, in this embodiment, the preset absolute value H0 of the slope of the forgetting curve predicted by the LSTM network is 1.2. The comparison process between the absolute value H of the slope of the forgetting curve predicted by the LSTM network and the preset absolute value H0 is as follows:

[0091] If the absolute value H of the slope of the forgetting curve predicted by the LSTM network is less than or equal to the preset absolute value H0, the number of qualified subjects pushed will be adjusted to 0.5 times the original number of subjects pushed. It should be noted that the adjusted number of qualified subjects pushed will be rounded up.

[0092] If the absolute value H of the slope of the forgetting curve predicted by the LSTM network is greater than the preset absolute value H0, the number of qualified subjects pushed will be adjusted to 0.9 times the original number of pushed subjects. It should be noted that the adjusted number of qualified subjects pushed will be rounded up.

[0093] Specifically, the analysis module described in this embodiment of the invention is also used to reduce the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage, and the difference is proportional to the reduction in the total number of questions pushed.

[0094] Specifically, in this embodiment, the preset difference I0 between the percentage of qualified subjects and the preset percentage is 0.1. The comparison process between the difference I0 and the preset difference I0 is as follows:

[0095] If the difference I between the percentage of qualified subjects and the preset percentage is less than or equal to the preset difference I0, the total number of questions pushed will be adjusted to 0.9 times the original total number of questions pushed. It should be noted that the adjusted total number of questions pushed will be rounded up.

[0096] If the difference I between the percentage of qualified subjects and the preset percentage is greater than the preset difference I0, the total number of questions pushed will be adjusted to 0.7 times the original total number of questions pushed. It should be noted that the adjusted total number of questions pushed will be rounded up.

[0097] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention. Example 1

[0098] In this embodiment, when the proportion of push notifications for one subject increases, the specific process for adjusting the proportion of push notifications for other subjects is as follows: Assuming there are three subjects: Chinese, Mathematics, and English, and the number of knowledge points involved in each subject is specifically in the order of Chinese > English > Mathematics, after adjusting the proportion of push notifications for Chinese from the initial value of 30% to 50%, the proportion of push notifications for Mathematics is adjusted first. The initial value of the proportion of push notifications for Mathematics is 25%. Normally, the proportion of push notifications for Mathematics should be adjusted from 25% to 5%. However, since the critical proportion is set at 10%, the proportion of push notifications for Mathematics is adjusted to 10%. Finally, the proportion of push notifications for English is adjusted from the initial value of 45% to 40%.

[0099] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0100] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent question-searching and adaptive recommendation system, characterized in that, include: The recognition module is used to recognize the questions entered by the user; An output module, connected to the recognition module, is used to acquire and output the answer corresponding to the question through big data. The statistics module, which is connected to the output module, is used to record and statistically analyze the search information during the search process, including the subject and knowledge points. The push module, which is connected to the statistics module, is used to push relevant test questions to the user based on the search information; The recording module, which is connected to the push module, is used to record the user's answer information when doing test questions, including the answering time and the score rate; An analysis module, connected to the recording module, is used to determine whether the push of test questions is qualified based on the obtained answer information, and to generate corresponding adjustment instructions based on the reasons for non-compliance, wherein the adjustment instructions include adjusting the percentage of qualified subjects and adjusting the total number of pushed questions. An adjustment module, connected to the analysis module, is used to adjust the corresponding parameter to the corresponding value based on the received adjustment command; The push module is also used to mark the proportion of each subject category in the question search within a preset time period as the proportion of each subject in the push process. The push module is also used to adjust the proportion of push notifications for each subject based on the number of knowledge point types in each subject; The push module is also used to increase the proportion of pushes for each subject based on the difference between the number of knowledge point types and the preset number, and the difference is proportional to the increase in the proportion of pushes. The push module is also used to adjust the push quantity ratio of each subject sequentially based on the adjusted push quantity ratio of a single subject to ensure that the sum of the push quantity ratios of each subject is 1. The push module is also used to determine the adjustment order of the proportion of pushes for each subject based on the ascending order of the number of knowledge point types contained in all search questions for each subject; The push module also has a critical push quantity ratio, and the adjusted push quantity ratio of each subject is greater than or equal to the critical push quantity ratio.

2. The intelligent question search and adaptive recommendation system according to claim 1, characterized in that, The analysis module is also used to determine whether the push notification for test questions is qualified based on the accuracy rate of a single subject, and to determine the reasons for the unqualified push notification for test questions based on the user's average answering time or the variance of the accuracy rate of a single knowledge point, wherein the accuracy rate of a single knowledge point is the answering accuracy rate of a user for all questions containing the same knowledge point under that subject.

3. The intelligent question search and adaptive recommendation system according to claim 2, characterized in that, The analysis module is also used to generate corresponding processing methods based on the comparison results between the user's average answering time and the preset answering time, including issuing a notification to adjust the proportion of push notifications, or adjusting the total number of push questions based on the difference between the average answering time and the preset answering time.

4. The intelligent question search and adaptive recommendation system according to claim 3, characterized in that, The analysis module is also used to reduce the total number of questions pushed based on the difference between the average answering time and the preset answering time, and the difference is inversely proportional to the reduction in the total number of questions pushed.

5. The intelligent question search and adaptive recommendation system according to claim 2, characterized in that, The analysis module is also used to generate corresponding processing methods based on the comparison results of the variance of the accuracy of a single knowledge point with the preset variance. This includes determining the reasons for the failure of the pushed questions based on the percentage of qualified subjects, or adjusting the number of related questions for each knowledge point based on the accuracy of each knowledge point. Qualified subjects are those whose variance of the accuracy of a single knowledge point is less than or equal to the preset variance.

6. The intelligent question search and adaptive recommendation system according to claim 5, characterized in that, The analysis module is also used to generate corresponding processing methods based on the comparison results between the percentage of qualified subjects and the preset percentage, including adjusting the number of qualified subjects pushed based on the absolute value of the slope of the forgetting curve predicted by the LSTM network, or adjusting the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage.

7. The intelligent question search and adaptive recommendation system according to claim 5, characterized in that, The analysis module is also used to reduce the number of qualified subjects pushed based on the absolute value of the slope of the forgetting curve predicted by the LSTM network, and the absolute value of the slope is inversely proportional to the reduction in the number of qualified subjects pushed.

8. The intelligent question search and adaptive recommendation system according to claim 5, characterized in that, The analysis module is also used to reduce the total number of questions pushed based on the difference between the percentage of qualified subjects and the preset percentage, and the difference is proportional to the reduction in the total number of questions pushed.

Images