Intelligent management system based on artificial intelligence

Abstract

The invention discloses an intelligent management system based on artificial intelligence, and relates to the technical field of artificial intelligence, and the system comprises a basic processing module which is used for collecting historical user information, and determining a corresponding initial association map; obtaining a corresponding target association map from the initial association map according to the current course; determining an initial association user according to the target user and the target association graph; the path determination module is used for determining a final course learning path corresponding to the target user in the current course according to the initial association user and the target association map; through combination of course scores and learning behavior data, differences among users are comprehensively reflected, and the accuracy of user distinguishing is improved; and the initial associated users are determined by using the total similarity, so that the individuation degree of the recommendation result is improved, and the learning requirements of different users are met.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to an intelligent management system based on artificial intelligence. Background Art

[0002] The continuous progress of artificial intelligence technology has promoted the wide application of intelligent education systems, covering multiple fields such as personalized learning, intelligent teaching assistance, and learning behavior analysis. However, when current technologies make learning plans for students, they only rely on the similarity between students for recommendations, so the effect of gradually improving students' abilities cannot be achieved.

[0003] The Chinese invention patent with the application number 202410826619.1 provides an intelligent education system based on artificial intelligence, which collects the course scores corresponding to associated courses of historical users, determines the first user difference of historical users in initial knowledge; determines the initial association graph; obtains the target association graph from the initial association graph according to the current course of the target user; determines the associated users of the target user in the target association graph according to the similarity, and then determines the course learning path.

[0004] Determine associated users and target courses according to course difficulty, user preferences, and learning progress. However, in actual applications, the learning habits, learning modes, and learning status changes of each user are different, resulting in vastly different effects when different users learn the same course. Therefore, in the existing patent technology, the specific learning status mode changes are not considered when determining the course learning path, and the real-time changes in the improvement of users' abilities cannot be tracked, reducing the effect of improving students' abilities. Summary of the Invention

[0005] This application provides an intelligent management system based on artificial intelligence, which solves the problem in the prior art that the learning status mode changes of users are not considered, and realizes the technical effect of accurately determining the learning plan according to individual difference factors and improving students' abilities.

[0006] This application provides an intelligent management system based on artificial intelligence, and the system includes: A basic processing module, configured to collect historical user information, determine a corresponding initial association graph according to the historical user information; obtain a target user and the current course corresponding to the target user, and obtain a corresponding target association graph from the initial association graph according to the current course; determine an initial associated user according to the target user and the target association graph; A path determination module, configured to determine the final course learning path corresponding to the target user in the current course according to the initial associated user and the target association graph; A course recommendation module, configured to obtain a course recommendation result corresponding to the target user according to the final course learning path; Wherein, the path determination module includes a forward path determination unit, a reverse path determination unit, and a final path determination unit; The forward path determination unit is configured to determine a forward course learning path corresponding to the target user in the current course according to the initial associated user and the target association graph; The reverse path determination unit is configured to determine an end associated user according to the initial associated user, determine multiple reverse demand paths according to the end associated user and the forward course learning path, and determine a target reverse demand path according to the multiple reverse demand paths; The final path determination unit is configured to obtain a final course learning path according to the forward course path and the target reverse demand path.

[0007] Further, the basic processing module includes an information collection unit, a difference calculation unit, a graph construction unit, a graph determination unit, and a user determination unit; Wherein, the information collection unit is configured to obtain associated courses corresponding to initial knowledge, collect course scores corresponding to the associated courses by historical users, and collect learning behavior data of historical users; The difference calculation unit is configured to determine a first user difference corresponding to the historical user in the initial knowledge according to the course scores, and determine a second user difference corresponding to the historical user in the initial knowledge according to the learning behavior data; The graph construction unit is configured to determine an initial association graph corresponding to the historical user according to the first user difference and the second user difference; The graph determination unit is configured to obtain a target user and a current course corresponding to the target user, and obtain a corresponding target association graph from the initial association graph according to the current course; The user determination unit is configured to calculate the total similarity between the target user and the user corresponding to the target association graph, and determine an initial associated user corresponding to the target user in the target association graph according to the total similarity.

[0008] Further, determining the forward course learning path corresponding to the target user in the current course according to the initial associated user and the target association graph includes: According to the position of the initial associated user corresponding to the target association graph, obtain the secondary associated users of the initial associated user, and combine the course contents corresponding to different secondary associated users in the current course to form the forward course learning path corresponding to the target user in the current course.

[0009] Further, determine the ending associated user according to the initial associated user, determine multiple reverse demand paths according to the ending associated user and the forward course learning path, and determine the target reverse demand path according to the multiple reverse demand paths, including: Obtain the historical learning path of the initial associated user, identify the stage period of the current course, mark the initial associated user at the stage period as the ending associated user, and obtain the achievement quantization value of the ending associated user as the target value of the target user; Obtain each course content in the forward course learning path, and mark them as relay nodes respectively, record the growth indicators of each relay node to form a growth indicator set, determine the target range of the target user in each relay node according to the target value and the growth indicator set, and connect the target ranges in each relay node to form multiple reverse demand paths; Wherein, the growth indicator refers to the factor affecting the learning achievement in the corresponding relay node, including learning duration and assessment score.

[0010] Further, determine the ending associated user according to the initial associated user, determine multiple reverse demand paths according to the ending associated user and the forward course learning path, and determine the target reverse demand path according to the multiple reverse demand paths, further including: Obtain the learning behavior data of the target user, calculate the fitness with the target range in the multiple reverse demand paths according to the learning behavior data of the target user and the growth indicator set, and take the one with the highest fitness as the target reverse demand path.

[0011] Further, the initial associated user and the ending associated user are the same user in two different states corresponding to the current course. The initial associated user refers to the user before starting to learn the current course, and the ending associated user refers to the user after finishing learning the current course.

[0012] Further, the achievement quantization value is a numerical value obtained by quantitatively evaluating the comprehensive performance of the ending associated user in the current course, and the stage period refers to the time point when the current course learning ends.

[0013] Further, the path determination module further includes: a path analysis unit, which is used to monitor the actual completion effect of each relay node in the final course learning path, and compare the actual learning achievement of the target user in the relay node with the corresponding target range; if the actual learning achievement exceeds the target range, it is determined to meet the standard, if the actual learning achievement does not exceed the target range, it is determined not to meet the standard, mark each relay node according to the comparison result, and perform regulation according to the differential processing mechanism; Wherein, the comparison result includes the number, position of the relay nodes that do not meet the standard and the corresponding actual learning achievements.

[0014] Further, the differential processing mechanism includes: obtaining the number of unqualified relay nodes and the corresponding actual learning achievements, calculating the knowledge similarity between each subsequent relay node and the unqualified relay nodes, marking it as the first knowledge similarity, presetting a first similarity threshold, and marking the relay nodes corresponding to the knowledge similarity greater than the first similarity threshold as auxiliary relay nodes; If the number of unqualified relay nodes is less than a preset first threshold, determine the target auxiliary relay nodes, and reset the target range of the target auxiliary relay nodes according to the actual learning achievements; If it is not less than the preset first threshold, determine whether the unqualified relay nodes are continuous. If they are continuous, generate a first auxiliary learning path based on all unqualified relay nodes and the corresponding auxiliary relay nodes; if they are not continuous, determine the knowledge similarity between the unqualified relay nodes, mark it as the second knowledge similarity, preset a second similarity threshold, and generate a second auxiliary learning path according to the comparison result between the second knowledge similarity and the second similarity threshold.

[0015] Further, if the number of unqualified relay nodes is less than a preset first threshold, determining the target auxiliary relay nodes includes: Calculating the knowledge similarity between the auxiliary relay nodes, marking it as the third knowledge similarity, presetting a third similarity threshold, screening out the auxiliary relay nodes whose third knowledge similarity is not greater than the third similarity threshold and marking them as preliminary selected auxiliary relay nodes. If the knowledge content of all preliminary selected auxiliary relay nodes completely includes the knowledge content of the unqualified relay nodes, mark the preliminary selected auxiliary relay nodes as target auxiliary relay nodes; if the knowledge content of all preliminary selected auxiliary relay nodes does not completely include the knowledge content of the unqualified relay nodes, re-determine the target auxiliary relay nodes.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages: By combining the course scores and learning behavior data, it more comprehensively reflects the differences between users, improves the accuracy of user differentiation; uses the total similarity to determine the initial associated users, improves the personalization degree of the recommendation results, and meets the learning needs of different users; constructs an initial association graph according to the comprehensive user difference index, and determines the final course learning path accordingly, achieving the effect of further improving students' abilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the architecture of an intelligent management system based on artificial intelligence in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant accompanying drawings. The preferred embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0020] Example 1: As Figure 1 shown, an intelligent management system based on artificial intelligence, the system includes: A basic processing module, configured to collect historical user information, determine a corresponding initial association graph according to the historical user information; obtain a target user and a current course corresponding to the target user, and obtain a corresponding target association graph from the initial association graph according to the current course; determine an initial associated user according to the target user and the target association graph.

[0021] In some embodiments, the basic processing module includes an information collection unit, a difference calculation unit, a graph construction unit, a graph determination unit, and a user determination unit.

[0022] Specifically, the information collection unit is configured to obtain an associated course corresponding to the initial knowledge, collect the course scores corresponding to the associated course by historical users, and collect the learning behavior data of historical users.

[0023] In some embodiments, the initial knowledge may be a knowledge point or passage corresponding to a certain subject. For example, the initial knowledge is the calculus knowledge point corresponding to the mathematics subject, or the initial knowledge is the grammar knowledge point corresponding to the English subject, etc. The associated course corresponding to the initial knowledge may be an online recorded video or a teaching text, etc.

[0024] In some embodiments, the learning behavior data includes, but is not limited to, learning time (such as daily learning duration, learning period preference), learning frequency (such as number of learning days per week, frequency of learning sessions), learning preference (such as preferred learning resource type, learning method, learning speed), and data reflecting learning attitude (such as homework completion rate, completion time).

[0025] Specifically, for the learning behavior data of users, the intelligent management system can automatically record the users' behaviors on the learning platform, such as login time, time spent on learning pages, resource access records, etc.; it can also conduct a questionnaire survey in the form of distributing electronic questionnaires, design questionnaires to collect the reported learning habits of users, including preferences, habits, and learning time periods, etc.; at the same time, analyze the interaction data between users and the system, including the effective learning time of users, answering time, and so on. Process the collected data, perform data cleaning, including removing outliers or extreme data points, such as a sudden surge in learning time may be caused by special events; establish a long-term observation period, establish a baseline of users' learning behaviors through long-term data collection, and reduce the influence of contingency; conduct multi-dimensional evaluation, combine multiple learning habit indicators to form a comprehensive understanding of users' learning habits, and avoid biases brought by single indicators.

[0026] A difference calculation unit, configured to determine a first user difference corresponding to the historical user in the initial knowledge according to the course score, and determine a second user difference corresponding to the historical user in the initial knowledge according to the learning behavior data.

[0027] In some embodiments, feature extraction is performed on the course score data to find the key factors affecting user differences, such as course difficulty, user preferences, learning progress, etc. Then, statistical methods or machine learning algorithms (such as clustering, principal component analysis, etc.) are used to calculate the differences in the course scores of historical users, find the differences between users, and further determine the measurement criteria for user differences, such as Euclidean distance, cosine similarity, etc., and determine the first user difference between historical users according to the measurement criteria.

[0028] In some embodiments, to determine the second user difference corresponding to the historical user in the initial knowledge according to the learning behavior data, determine the key learning indicators of each user, extract the contained learning behavior data from the specific behaviors of the users, and quantify the learning behavior data respectively. For example, convert the learning time into a learning duration index, convert the learning frequency into a learning activity index, etc. According to the quantified indicators, determine the key learning indicators that best fit the user, that is, the indicator data that can best reflect the user's learning behavior, and set the corresponding weight values. For example, the learning time and learning frequency of User A are the most regular and can best reflect the user's learning behavior. Among all the learning behavior data of User A, set the weight values of learning time and learning frequency to larger values (such as 0.4, 0.4), and then set the weight values of the remaining learning behavior data.

[0029] A graph construction unit, configured to determine an initial association graph corresponding to the historical user according to the first user difference and the second user difference.

[0030] In some embodiments, the graph construction unit determines historical users as nodes in the initial association graph, and then connects the historical users. The association relationships between historical users are connected through the first user difference and the second user difference, and the direction between historical users is determined according to the magnitudes of the course scores corresponding to the historical users, with the user having a lower course score pointing to the user having a higher course score, thereby obtaining the initial association graph; the first user difference and the second user difference are weighted and combined to form a comprehensive user difference index, and the weights need to be preset according to the actual situation. A graph theory algorithm (such as minimum spanning tree, K-nearest neighbor graph, etc.) is used to construct the initial association graph according to the comprehensive difference index, where the nodes represent users and the edges represent the comprehensive difference index between users.

[0031] The graph determination unit is configured to obtain a target user and the current course corresponding to the target user, and obtain the corresponding target association graph from the initial association graph according to the current course.

[0032] In some embodiments, the current course corresponding to the target user is determined through the historical learning records, personal profile, or real-time user behavior data of the target user, and then the similarity between the current course and the associated courses is calculated. Furthermore, the target association graph corresponding to the current course is filtered out from the initial association graph according to the relevant courses.

[0033] The user determination unit is configured to calculate the total similarity between the target user and the users corresponding to the target association graph, and determine the initial associated user corresponding to the target user in the target association graph according to the total similarity.

[0034] In some embodiments, similarity measurement methods (such as cosine similarity, Pearson correlation coefficient, etc.) are used to calculate the similarity between the target user and each user in the target association graph. This requires using user characteristics, learning behaviors, or other relevant data to compare the similarity between users. Find the highest value among the calculated similarities, which is the total similarity. The total similarity reflects the highest similarity degree between the target user and a certain user in the target association graph, and the highest similarity degree is obtained by calculating the similarity degrees in two aspects of the first user difference and the second user difference. Thus, the initial associated user corresponding to the target user in the target association graph is determined according to the total similarity. This step requires mapping the similarity calculation results to the corresponding users to find the user most similar to the target user.

[0035] The path determination module is configured to determine the final course learning path corresponding to the target user in the current course according to the initial associated user and the target association graph; The course recommendation module is configured to obtain the course recommendation result corresponding to the target user according to the final course learning path; Specifically, according to the final course learning path, the learning videos corresponding to each stage are obtained, and then the similar videos corresponding to each learning video are obtained, so as to calculate the degree of fit between each similar video and the target user, select the target video corresponding to the target user at each stage, and obtain the course recommendation result corresponding to the target user by combining the target videos in sequence.

[0036] The technical solutions in the embodiments of the present application at least have the following technical effects or advantages: By combining the course scores and learning behavior data, the present application more comprehensively reflects the differences between users, improves the accuracy of user differentiation; uses the total similarity to determine the initial associated users, improves the personalization degree of the recommendation results, and meets the learning needs of different users; constructs the initial associated graph according to the comprehensive user difference index, and determines the final course learning path based on this, achieving the effect of further improving the students' abilities.

[0037] Embodiment 2: In Embodiment 1, the learning behavior data of the user is added to ensure that the generated learning path can be suitable for the target user and improve the learning effect. This embodiment makes further improvements on the basis of the above embodiment.

[0038] The path determination module includes a forward path determination unit, a reverse path determination unit, and a final path determination unit. Among them, the forward path determination unit is used to determine the forward course learning path corresponding to the target user in the current course according to the initial associated user and the target associated graph; The reverse path determination unit is used to determine the end associated user according to the initial associated user, determine multiple reverse demand paths according to the end associated user and the forward course learning path, and determine the target reverse demand path according to the multiple reverse demand paths; The final path determination unit is used to obtain the final course learning path according to the forward course path and the target reverse demand path; Specifically, determining the forward course learning path corresponding to the target user in the current course according to the initial associated user and the target associated graph includes: obtaining the secondary associated users of the initial associated user according to the position of the initial associated user corresponding to the target associated graph, and combining the course contents corresponding to the different secondary associated users in the current course to form the forward course learning path corresponding to the target user in the current course.

[0039] In some embodiments, according to the position of the initial associated user corresponding to the target association graph, obtain the secondary associated users that the initial associated user needs to rely on when being promoted in sequence from the association relationships in the target association graph, and then obtain the learning videos of the secondary associated users for the current course. The course content between different secondary associated users forms the corresponding positive course learning path for the target user in the current course. The above positive course learning path includes the entire process of the target user learning the current course, and the course content includes relevant videos or other materials that are helpful for the learning process. The relevant videos or other materials here include all the learning content that can provide help among the initial associated users and the secondary associated users. At the same time, a screening mechanism is preset to delete learning videos with high repetition to avoid redundant learning content.

[0040] In some embodiments, the number of the positive course learning paths is at least one, and the positive course learning paths and the reverse demand paths are in one-to-one correspondence. If there are multiple positive course learning paths and reverse demand paths, determine the optimal reverse demand path according to the learning behavior data of the target user, and mark it as the target reverse demand path. Obtain the final course learning path according to the target reverse demand path and the corresponding positive course learning path.

[0041] Determine the ending associated user according to the initial associated user, determine multiple reverse demand paths according to the ending associated user and the positive course learning path, and determine the target reverse demand path according to the multiple reverse demand paths, including: Obtain the historical learning path of the initial associated user, identify the stage end point of the current course, mark the initial associated user at the stage end point as the ending associated user, and obtain the achievement quantification value of the ending associated user as the target value of the target user; obtain each course content in the positive course learning path, and mark them as relay nodes respectively, record the growth indicators of each relay node to form a growth indicator set, determine the target range of the target user in each relay node according to the target value and the growth indicator set, and connect the target ranges in each relay node to form multiple reverse demand paths; Wherein, the growth indicator refers to the factors that affect the learning achievement in the corresponding relay node, including learning duration and assessment score.

[0042] In some embodiments, the positive course learning path is composed of several learning videos or other learning content, and the relay node represents each course content in the current course; the growth indicator of each relay node is the specific requirement of the corresponding course content, and this specific growth requirement is inferred reversely according to the historical learning path of the ending associated user. For example, the completion degree of the video is above 80% and the assessment score is above 80 points, etc. The growth indicator set refers to the set of all growth indicators.

[0043] In some embodiments, the achievement quantification value is a value obtained by quantitatively evaluating the comprehensive performance of the end-associated user in the current course. Specifically, it is calculated by weighted summation through comprehensively considering multiple dimensions such as test scores, homework completion, learning activity, and learning duration. The homework completion is scored according to the submission situation of the user's homework (such as whether it is submitted on time, the number of submissions, etc.), the completion quality (such as accuracy, integrity, innovation, etc.), and the homework difficulty. The scoring can adopt a percentage system, a grading system, or a combination of specific scores and comments; the learning activity is given extra points according to the activity of the user in the course discussion area (such as the number of posts, the number of replies, etc.), the contribution (such as providing valuable information, solving others' problems, etc.), and the discussion quality; the activity can be evaluated by counting the number of posts and replies of the user, and the contribution can be reflected by the number of times the user is liked, cited, or rated as a high-quality post; the learning duration is given corresponding extra points or grade evaluations according to the total duration of the user's learning of this course, combined with the difficulty and importance of the course content, and can also be automatically recorded by the learning management system or the online learning platform. The scores of the above aspects are weighted and summed according to the preset weights to obtain the achievement quantification value of the user. The weights should be reasonably set according to the characteristics of the current course, the teaching objectives, and the importance of each dimension in the evaluation, and ensure the fairness and transparency of the evaluation process.

[0044] The stage period refers to the time point when the learning of the current course ends. Specifically, the learning progress of the user is monitored by the intelligent management system. When the user reaches the predetermined learning duration of a certain stage and completes all the learning tasks of that stage, it can be identified as the stage period. At the same time, the user is allowed to submit stage completion feedback during the learning process. When the system receives the user's stage completion application and after verification and confirmation, the stage period can be determined. When the user passes the stage test or evaluation and completes all the learning tasks of that stage, it is identified as the stage period, and the user is marked as the end-associated user.

[0045] In some embodiments, the growth indicator refers to the factors that affect learning achievements in the corresponding relay node, including but not limited to learning duration, assessment scores, completion progress, and learning activity, etc.; the stage period refers to the time point when the learning of the current course ends.

[0046] Determine the end-associated user according to the initial associated user, determine multiple reverse demand paths according to the end-associated user and the forward course learning path, and determine the target reverse demand path according to the multiple reverse demand paths. It also includes: Obtain the learning behavior data of the target user, calculate the fitness with the target range in the multiple reverse demand paths according to the learning behavior data of the target user and the growth indicator set, and take the one with the highest fitness as the target reverse demand path.

[0047] In some embodiments, the target range refers to the lowest target range that the target user wants to reach the target value, that is, the learning achievement standard that the user needs to achieve in each relay node. The target range is determined according to the following two aspects: analyze the growth indicators in each relay node, such as learning duration, assessment score, completion progress, etc., to determine the lowest standard that the user needs to achieve in this node; refer to the learning achievement data of historical users in this node to determine the target range.

[0048] Specifically, conduct a detailed analysis of the growth indicators in each relay node to determine the key factors affecting learning achievements; set the target range for each relay node according to the growth indicator analysis and course requirements.

[0049] Preferably, the reverse demand path sets corresponding growth indicators in the target range corresponding to each relay node. It is not difficult to understand that a forward course learning path includes several relay nodes, and the mastery level of each relay node is different. There is a correlation between each relay node. If the learning situation in a relay node does not reach the corresponding target range, and if it exceeds the original target range in the next relay node, the effects that these two relay nodes want to achieve are also realized; the target range also represents the knowledge content that the target user should master in this relay node.

[0050] In some embodiments, according to the target value and the growth indicator set, determine the target range of the target user in each relay node, which can connect to form multiple reverse demand paths. Each reverse demand path corresponds to a different target range. Obtain the learning behavior data of the target user. According to the learning behavior data of the target user and the growth indicator set, calculate the fitness with the target range in multiple reverse demand paths, and take the one with the highest fitness as the target reverse demand path, that is, the reverse demand path most suitable for the target user. Combine the target reverse demand path with the forward course path to form the final course learning path. Specifically, the fitness needs to be calculated according to the actual learning behavior data of the target user. Judge its fitness according to the gap between the target range and the current target user. Judge the gap with the target range according to the current learning behavior data of the target user, and then obtain the fitness. The target range includes multiple learning indicators, and the completion difficulty of each learning indicator is different for the target user. Initially judge the difficulty value of each learning indicator in each target range, normalize all the data, and use the method of weighted summation to obtain the final fitness. The greater the difficulty, the greater the gap, and the smaller the fitness. It is necessary to conform to the law of the user's learning growth so that progress can be made, and at the same time, the enthusiasm of the target user should be improved.

[0051] The initial associated user and the end associated user are the same user in two different states corresponding to the current course. The initial associated user refers to the user when the current course has not started learning, and the end associated user refers to the user when the current course learning is completed.

[0052] In some embodiments, a target range is obtained from the target reverse demand path, and the forward course path in which each relay node is marked with the corresponding target range is used as the final course learning path. The final course learning path not only includes the learning courses necessary for the target user to learn the initial knowledge points, but also includes the specific indicators and requirements of each relay node, thus ensuring the actual growth effect of the target user.

[0053] The technical solutions in the embodiments of the present application at least have the following technical effects or advantages: The present application sets a reverse demand path, and comprehensively generates a final course learning path according to the forward course learning path and the target reverse demand path, determines the corresponding standards that the target user needs to achieve when using the final course learning path, realizes the effect of further accurately generating a personalized learning path, and clearly quantifies the learning achievements of the target user.

[0054] Embodiment 3: In the above embodiment, a reverse demand path is set, and a final course learning path is comprehensively generated according to the forward course learning path and the target reverse demand path. However, the actual effects corresponding to different users are not the same, and the actual effects at each relay node cannot reach the preset target range. This embodiment makes further improvements on the basis of the above embodiment.

[0055] The path determination module further includes: a path analysis unit, configured to monitor the actual completion effect of each relay node in the final course learning path, compare the actual learning achievements of the target user at the relay node with the corresponding target range; if the actual learning achievements exceed the target range, it is determined to meet the standard, if the actual learning achievements do not exceed the target range, it is determined not to meet the standard, mark each relay node according to the comparison result, and perform regulation according to the differential processing mechanism; Among them, the comparison result includes the number, position and corresponding actual learning achievements of the relay nodes that do not meet the standard.

[0056] In some embodiments, the actual completion effect refers to the actual learning progress or achievements of the target user at the relay node, which can be the scores or completion situations obtained through tests, homework or other evaluation methods. The comparison result is obtained by separately comparing each index in the actual completion effect with each index in the target range. If all indexes are greater than the indexes in the target range, it is determined to meet the standard, otherwise it is determined not to meet the standard.

[0057] The differential processing mechanism includes: obtaining the number of unqualified relay nodes and the corresponding actual learning achievements, calculating the knowledge similarity between each subsequent relay node and the unqualified relay nodes, marking it as the first knowledge similarity, presetting a first similarity threshold, and marking the relay nodes corresponding to the knowledge similarity greater than the first similarity threshold as auxiliary relay nodes; If the number of unqualified relay nodes is less than a preset first threshold, determine the target auxiliary relay node and reset the target range of the target auxiliary relay node according to the actual learning achievements; If it is not less than the preset first threshold, determine whether the unqualified relay nodes are continuous. If they are continuous, generate a first auxiliary learning path based on all the unqualified relay nodes and the corresponding auxiliary relay nodes; if they are not continuous, determine the knowledge similarity between the unqualified relay nodes, mark it as the second knowledge similarity, preset a second similarity threshold, and generate a second auxiliary learning path according to the comparison result between the second knowledge similarity and the second similarity threshold.

[0058] In some embodiments, the differential processing mechanism is a strategy for dynamically adjusting the learning path or providing additional assistance according to the user's learning achievements on the relay nodes. It is necessary to match the processing method according to the specific actual situation. First, determine the number of unqualified relay nodes. The first knowledge similarity refers to the knowledge similarity between each subsequent relay node and the unqualified relay nodes, which is used to determine whether there is the same knowledge as the unqualified relay nodes in the subsequent relay nodes. If so, further determine the similarity degree, compare it with the preset first similarity threshold, and select the auxiliary relay node. It is easy to understand that initially determining the unqualified relay nodes means that the target user has not fully mastered the knowledge content here, that is, has not reached the preset standard. If there are subsequent relay nodes with the same knowledge, the subsequent standard needs to be reset to improve the corresponding learning effect. The first similarity threshold, the second similarity threshold, and the first threshold are all preset according to the actual situation or historical data experience, and this application does not make specific limitations. In this application, the calculation of similarity can adopt various similarity measurement calculation methods (such as cosine similarity), which are well-known technologies to those skilled in the art, and this application does not make specific limitations.

[0059] The first knowledge similarity is calculated based on the number of knowledge contents of the unqualified relay nodes contained in the subsequent relay nodes. Specifically, if the subsequent relay node includes all the knowledge contents of the unqualified relay node, the knowledge similarity is 100%; if the subsequent relay node includes 50% of the knowledge contents of the unqualified relay node, the knowledge similarity is 50%.

[0060] The second knowledge similarity refers to the knowledge similarity between discontinuous unqualified relay nodes. By calculating the similarity, the specific range of the target user's knowledge gap can be judged, and it can be determined whether the target user has difficulty in understanding specific subjects or concepts.

[0061] In some embodiments, when it is judged that there are continuous unqualified relay nodes, a first auxiliary learning path is generated according to all unqualified relay nodes and corresponding auxiliary relay nodes. The first auxiliary learning path includes the following setting methods: judge the distance between the unqualified relay node and the corresponding auxiliary relay node. If the distance exceeds three relay nodes, further judge whether there is an association between the knowledge content of the remaining intermediate relay nodes and the unqualified relay node. If there is an association, generate a first auxiliary learning path according to the auxiliary relay node, and set the first auxiliary learning path before the remaining intermediate relay nodes for learning; if there is no association, modify the target range of the corresponding auxiliary relay node.

[0062] Preferably, the first auxiliary learning path can only select the content of the auxiliary relay node, or can also select other learning content.

[0063] In some embodiments, if they are discontinuous, judge the knowledge similarity between unqualified relay nodes, mark it as the second knowledge similarity, preset a second similarity threshold, and generate a second auxiliary learning path according to the comparison result between the second knowledge similarity and the second similarity threshold.

[0064] In some embodiments, for the discontinuous situation, there are jump-type unqualified relay nodes, and the range of knowledge points involved is relatively large. According to the comparison result between the second knowledge similarity and the second similarity threshold, judge the knowledge similarity between unqualified relay nodes, and generate a second auxiliary learning path according to the associated knowledge content as additional auxiliary learning. Further solve the problems of inconsistent learning results of users at different relay nodes and inaccurate processing of unqualified nodes. The technical solutions in the above embodiments of the present application at least have the following technical effects or advantages: By using the path analysis unit to monitor the actual completion effect of the relay node and combining the differential processing mechanism to dynamically adjust the learning path or provide an auxiliary learning path, the present application achieves the effect of personalized adjustment of the course learning path according to the user's learning results.

[0065] Embodiment 4: In the above embodiment, a one-to-one comparison method is used to judge unqualified relay nodes and auxiliary relay nodes. For decentralized knowledge content, it is easy to cause a low similarity situation. This embodiment makes further improvements on the basis of the above embodiment.

[0066] If the number of non-compliant relay nodes is less than a preset first threshold, determining target auxiliary relay nodes includes: Calculating the knowledge similarity between the auxiliary relay nodes, marked as the third knowledge similarity, presetting a third similarity threshold, screening out the auxiliary relay nodes whose third knowledge similarity is not greater than the third similarity threshold and marking them as preliminary selected auxiliary relay nodes. If the knowledge content of all the preliminary selected auxiliary relay nodes completely includes the knowledge content of the non-compliant relay nodes, then marking the preliminary selected auxiliary relay nodes as target auxiliary relay nodes; if the knowledge content of all the preliminary selected auxiliary relay nodes does not completely include the knowledge content of the non-compliant relay nodes, then re-determining the target auxiliary relay nodes; wherein, the number of the target auxiliary relay nodes is multiple.

[0067] In some embodiments, further determining the knowledge similarity between the auxiliary relay nodes, deleting the auxiliary relay nodes with very high similarity, selecting auxiliary relay nodes with different contents, and judging whether the knowledge content covered by the selected auxiliary relay nodes completely includes all the knowledge content of the non-compliant relay nodes. If all are included, then marking the preliminary selected auxiliary relay nodes as target auxiliary relay nodes, and re-determining the target range according to the selected target auxiliary relay nodes to improve the learning effect of the target user. The target auxiliary relay nodes refer to those auxiliary relay nodes that are selected to help users make up for the knowledge gaps of the non-compliant relay nodes. The number of them is multiple, and their knowledge content should completely include the knowledge content of the non-compliant relay nodes. The third similarity threshold is a preset value used to judge whether the knowledge similarity between the auxiliary relay nodes is low enough to ensure the diversity of the selected auxiliary relay nodes in terms of content. It is preset according to the actual situation or historical data experience, and the present application does not make specific restrictions.

[0068] The technical solutions in the embodiments of the present application above have at least the following technical effects or advantages: The present application calculates the third knowledge similarity between the auxiliary relay nodes, sets the third similarity threshold to screen the preliminary selected auxiliary relay nodes, judges whether the knowledge content of the preliminary selected auxiliary relay nodes completely covers the knowledge content of the non-compliant relay nodes, and optimizes the selection of the target auxiliary relay nodes, thus achieving the effect of further improving the learning improvement of the target user.

[0069] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent management system based on artificial intelligence, characterized in that: The system comprises: A basic processing module is used to collect historical user information, determine the corresponding initial association map according to the historical user information; obtain a target user and a current course corresponding to the target user, obtain a corresponding target association map from the initial association map according to the current course; determine the initial association user according to the target user and the target association map; A path determination module, used to determine a final course learning path corresponding to the target user in the current course according to the initial associated user and the target associated graph; A course recommendation module, used to obtain a course recommendation result corresponding to the target user according to the final course learning path; Wherein, the path determination module includes a forward path determination unit, a reverse path determination unit and a final path determination unit; The forward path determination unit is used to determine the forward course learning path corresponding to the target user in the current course according to the initial associated user and the target association map; The reverse path determination unit is used to determine the ending association user according to the initial association user, determine multiple reverse demand paths according to the ending association user and the forward course learning path, and determine a target reverse demand path according to the multiple reverse demand paths; The final path determination unit is used to obtain a final course learning path according to the forward course path and the target reverse demand path.

2. The intelligent management system based on artificial intelligence as claimed in claim 1, characterized in that: The basic processing module includes an information collection unit, a difference calculation unit, a graph construction unit, a graph determination unit and a user determination unit; The information collection unit is used to obtain the associated courses corresponding to the initial knowledge, collect the course scores of historical users corresponding to the associated courses, and collect the learning behavior data of historical users; The difference calculation unit is used to determine the first user difference corresponding to the historical user in the initial knowledge according to the course score, and determine the second user difference corresponding to the historical user in the initial knowledge according to the learning behavior data; The graph construction unit is used to determine the initial association graph corresponding to the historical user according to the first user difference and the second user difference; The graph determination unit is used to obtain a target user and a current course corresponding to the target user, and obtain a corresponding target association graph from the initial association graph according to the current course; The user determination unit is used to calculate the total similarity between the target user and the users corresponding to the target association graph, and determine the initial associated users corresponding to the target user in the target association graph according to the total similarity.

3. The intelligent management system based on artificial intelligence as claimed in claim 1, characterized in that: Determining a positive course learning path corresponding to the target user in the current course according to the initial associated user and the target associated graph includes: According to the position of the initial associated user in the target associated graph, the secondary associated users of the initial associated user are obtained, and the course contents corresponding to different secondary associated users in the current course are combined to form a positive course learning path corresponding to the target user in the current course.

4. The intelligent management system based on artificial intelligence as claimed in claim 1, characterized in that: Determining an end-association user according to the initial association user, determining a plurality of reverse demand paths according to the end-association user and the forward course learning path, and determining a target reverse demand path according to the plurality of reverse demand paths, including: Acquire the historical learning path of the initial associated user, identify the stage period of the current course, mark the initial associated user at the stage period as the ending associated user, and acquire the achievement quantified value of the ending associated user as the target value of the target user; Obtain each course content in the forward course learning path, and mark them as relay nodes respectively, record the growth index of each relay node to form a growth index set, determine the target range of the target user in each relay node according to the target value and the growth index set, and connect the target range in each relay node to form multiple reverse demand paths; The growth indicators refer to factors that affect learning outcomes in the corresponding relay nodes, including learning time and evaluation scores.

5. The intelligent management system based on artificial intelligence as claimed in claim 1, characterized in that: Determining the end-association user according to the initial associated user, determining a plurality of reverse demand paths according to the end-association user and the forward course learning path, and determining a target reverse demand path according to the plurality of reverse demand paths, further comprising: The learning behavior data of the target user is obtained, and the degree of fit with the target range in multiple reverse demand paths is calculated based on the learning behavior data and growth indicator set of the target user, and the one with the highest degree of fit is taken as the target reverse demand path.

6. The intelligent management system based on artificial intelligence as claimed in claim 4, characterized in that: The initial associated user and the final associated user are the same user in two different states corresponding to the current course. The initial associated user refers to the user before the current course begins, and the final associated user refers to the user after the current course ends.

7. The intelligent management system based on artificial intelligence as claimed in claim 4, characterized in that: The achievement quantification value is a numerical value that quantifies the total performance of the associated user in the current course, and the stage period refers to the time point when the current course learning ends.

8. The intelligent management system based on artificial intelligence as claimed in claim 1, characterized in that: The path determination module also includes: a path analysis unit, which is used to monitor the actual completion effect of each relay node in the final course learning path, and compare the actual learning results of the target user at the relay node with the corresponding target range; if the actual learning results exceed the target range, it is determined to be up to standard; if the actual learning results do not exceed the target range, it is determined to be not up to standard, and each relay node is marked according to the comparison result, and regulated according to the differential processing mechanism; The comparison results include the number and location of the relay nodes that do not meet the standards and the corresponding actual learning outcomes.

9. The intelligent management system based on artificial intelligence as claimed in claim 8, characterized in that: The difference processing mechanism includes: obtaining the number of unqualified relay nodes and the corresponding actual learning results, calculating the knowledge similarity between each subsequent relay node and the unqualified relay node, marking it as a first knowledge similarity, presetting a first similarity threshold, and marking the relay nodes corresponding to the knowledge similarity greater than the first similarity threshold as auxiliary relay nodes; If the number of unqualified relay nodes is less than a preset first threshold, a target auxiliary relay node is determined, and a target range of the target auxiliary relay node is reset according to actual learning results; If it is not less than the preset first threshold, determine whether the non-compliant relay nodes are continuous. If they are continuous, generate a first auxiliary learning path based on all non-compliant relay nodes and corresponding auxiliary relay nodes. If they are not continuous, determine the knowledge similarity between the non-compliant relay nodes, mark it as the second knowledge similarity, pre-set the second similarity threshold, and generate a second auxiliary learning path based on the comparison result of the second knowledge similarity and the second similarity threshold.

10. The intelligent management system based on artificial intelligence as claimed in claim 9, characterized in that: If the number of unqualified relay nodes is less than a preset first threshold, determining a target auxiliary relay node includes: The knowledge similarity between the auxiliary relay nodes is calculated and marked as the third knowledge similarity, a third similarity threshold is pre-set, and the auxiliary relay nodes whose third knowledge similarity is not greater than the third similarity threshold are screened out and marked as preliminary auxiliary relay nodes. If the knowledge contents of all preliminary auxiliary relay nodes all include the knowledge contents of the unqualified relay nodes, the preliminary auxiliary relay nodes are marked as target auxiliary relay nodes; if the knowledge contents of all preliminary auxiliary relay nodes do not all include the knowledge contents of the unqualified relay nodes, the target auxiliary relay nodes are re-determined.

Images