A course recommendation method based on learning trajectory and knowledge graph

Abstract

The present invention discloses a course recommendation method based on learning trajectory and knowledge graph, which specifically relates to the technical field of educational course recommendation, and comprises the following steps: by collecting and integrating students' learning trajectory data, mapping to the knowledge graph to generate a learning trajectory-knowledge joint graph, performing knowledge point mastery and learning rate analysis, and obtaining the knowledge mastery value and learning rate of each knowledge point; based on the knowledge mastery value, the system divides the knowledge points into strengths and weaknesses; for weak knowledge points, the weak recommendation degree is calculated in combination with the knowledge mastery value and the learning rate, and courses are recommended preferentially according to the recommendation degree; for strong knowledge points, the strong recommendation degree is calculated based on the learning rate, and advanced courses are recommended to improve students' learning effect and personalized learning experience; the present invention can accurately identify students' strong and weak knowledge points, thereby providing highly targeted learning recommendations, so that students can learn more efficiently on a personalized learning path and improve learning outcomes.

Description

Technical Field

[0001] The present invention relates to the technical field of educational course recommendation, and more specifically, to a course recommendation method based on learning trajectory and knowledge graph. Background Art

[0002] Course recommendation based on learning trajectory and knowledge graph is an intelligent education recommendation method, which combines students' learning history data and knowledge graph to recommend courses in a personalized way. First of all, learning trajectory refers to the behavioral record left by students during the learning process, including learning progress, learning content, test scores, etc. By analyzing these data, we can understand students' learning habits and knowledge mastery. The knowledge graph is a network that represents the relationship between knowledge points. It connects each knowledge point through logic to help reveal which knowledge points students are deficient in or need further learning. By mapping students' learning trajectories to the knowledge graph, we can discover students' learning weaknesses and knowledge areas that need to be improved.

[0003] With the rapid development of educational technology, personalized learning has gradually become an important trend in online education. Traditional teaching methods often cannot accurately identify each student's knowledge and learning needs, resulting in rough course recommendations that are difficult to meet the personalized learning needs of different students. Therefore, a course recommendation method based on learning trajectory and knowledge graph is proposed. Summary of the invention

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A course recommendation method based on learning trajectory and knowledge graph includes the following steps:

[0006] Collect and integrate students' learning trajectory data to obtain an integrated set of students' learning trajectory data;

[0007] Construct a knowledge graph, decompose the course content into knowledge points, and then map the integrated set of students' learning trajectory data onto the knowledge graph to obtain a learning trajectory-knowledge joint graph;

[0008] Based on the learning trajectory-knowledge joint graph, the knowledge point mastery analysis is performed to obtain the knowledge mastery value of each knowledge point. At the same time, the knowledge point learning analysis is performed to obtain the knowledge learning rate of each knowledge point.

[0009] Based on the knowledge mastery value, the mastery degree of knowledge points is judged, and all knowledge points are divided into a strong knowledge point set and a weak knowledge point set;

[0010] For a set of weak knowledge points, the weak recommendation degree is obtained according to the knowledge mastery value and knowledge learning rate, and courses are recommended based on the weak recommendation degree;

[0011] For a set of strong knowledge points, the knowledge learning rate of the target knowledge point is the strong recommendation degree, and course recommendations are made based on the strong recommendation degree.

[0012] In a preferred embodiment, the knowledge point mastery analysis is performed based on the learning trajectory-knowledge joint graph, and the knowledge mastery value of each knowledge point is obtained as follows:

[0013] Get the total learning time that students spend on target knowledge point i and mark it as , the frequency of students’ review on target knowledge point i and marked as , the correct rate of the student's jth exercise on the target knowledge point i and marked as , the student's test score for the kth test on target knowledge point i and marked as ; Then substitute them into the knowledge mastery value calculation formula of target knowledge point i:

[0014] ; , , are the preset influence coefficients and , , The sum of is one, is a positive value, is the slope parameter of the logistic function, ranging from 0.01 to 10, Review frequency The corresponding preset threshold; n is the total number of exercises, m is the total number of exams, The preset adjustment coefficient for the jth exercise, is the adjustment coefficient for the kth test, is the knowledge mastery value of target knowledge point i.

[0015] In a preferred embodiment, a student learning model is established based on the knowledge mastery value:

[0016] Get the student's knowledge mastery value at the starting time point of the target knowledge point i and mark it as , the maximum mastery value increment that students can eventually achieve and marked as , build the following model:

[0017] ; to minimize the predicted value The difference with the actual value is used to determine the attenuation coefficient The numerical value of represents the predicted knowledge mastery value corresponding to time t, and the attenuation coefficient The determined values ​​are substituted into the student learning model for application.

[0018] In a preferred embodiment, the knowledge point learning analysis is performed to obtain the knowledge learning rate of each knowledge point, which refers to:

[0019] Get the student learning model applied by the student on the target knowledge point i, and then the student's knowledge learning rate on the target knowledge point i:

[0020] ; Represents the student's knowledge learning rate on target knowledge point i.

[0021] In a preferred embodiment, judging the mastery degree of a knowledge point based on the knowledge mastery value refers to:

[0022] The student's knowledge mastery value on each knowledge point is obtained respectively and compared with the preset knowledge point mastery threshold. If the knowledge mastery value on the knowledge point is greater than or equal to the preset knowledge point mastery threshold, the knowledge point is marked as a learning strength knowledge point. If the knowledge mastery value on the knowledge point is less than the preset knowledge point mastery threshold, the knowledge point is marked as a learning weakness knowledge point.

[0023] In a preferred embodiment, dividing all knowledge points into a strong knowledge point set and a weak knowledge point set means:

[0024] All knowledge points marked as learning strengths are summarized to obtain a set of strengths knowledge points; all knowledge points marked as learning weaknesses are summarized to obtain a set of weaknesses knowledge points.

[0025] In a preferred embodiment, obtaining the weak point recommendation degree according to the knowledge mastery value and the knowledge learning rate means:

[0026] Map the knowledge mastery value and knowledge learning rate to the numerical range of [0,1] respectively to obtain the mapped knowledge mastery value , Knowledge Learning Rate , and then substitute it into the weak recommendation calculation formula:

[0027] ; Recommended for weak points, , Both are preset recommendation coefficients and their sum is one.

[0028] In a preferred embodiment, recommending courses based on the degree of recommendation of weaknesses refers to:

[0029] Recommend weak points Arrange in descending order to obtain the weakness recommendation main page list. Click on each knowledge point on the weakness recommendation main page list to enter the weakness recommendation sub-page list corresponding to the knowledge point. The weakness recommendation sub-page list is composed of multiple courses containing the knowledge point, and the multiple courses containing the knowledge point are arranged in descending order according to the content similarity between the course and the knowledge point.

[0030] Technical effects and advantages of the present invention:

[0031] By analyzing students' learning trajectories and knowledge graphs, the present invention can customize personalized learning paths for students. According to students' knowledge mastery values ​​and learning rates, students' strengths and weaknesses can be accurately identified, thereby providing highly targeted learning recommendations, allowing students to learn more efficiently on personalized learning paths and improve learning outcomes.

[0032] The present invention can dynamically adjust the recommended learning content and course difficulty, and provide the most suitable learning materials through comprehensive analysis of learning trajectory, knowledge mastery value and learning rate. This not only improves learning efficiency, but also helps students master knowledge faster, and further consolidates and improves strengths and strengthens weaknesses. By continuously updating and tracking students' learning trajectories, the present invention can provide real-time feedback on students' learning status. Both bottlenecks in learning and progress can be identified and adjusted in a timely manner, helping students dynamically optimize their personal learning strategies.

[0033] The present invention can accurately evaluate students' weak and strong knowledge points through comprehensive analysis of knowledge mastery value and knowledge learning rate. For weak knowledge points, the system can give priority to recommending related courses to help students make up for knowledge gaps; for strong knowledge points, the system recommends higher-level advanced courses to further strengthen students' knowledge reserves and application capabilities in these areas. The present invention can reasonably allocate learning resources and avoid unnecessary waste of resources through intelligent knowledge point analysis and course recommendations. Students' time and energy can be concentrated on the areas that need improvement the most, maximizing learning results. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0035] Figure 1 This is a schematic diagram of a course recommendation method based on learning trajectory and knowledge graph in the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] Reference Figure 1 The following examples are obtained:

[0038] Embodiment 1: A course recommendation method based on learning trajectory and knowledge graph, comprising the following steps:

[0039] Collect and integrate students' learning trajectory data to obtain the integrated set of students' learning trajectory data; learning trajectory data reflects students' learning behavior and learning process, and is the basis for understanding students' learning habits and knowledge mastery. These data can include students' learning time, number of exercises, accuracy, test scores, etc. Learning trajectory data can come from different learning systems and platforms. The integration process is not limited to simple data merging, but also requires data cleaning and normalization. The learning behaviors that can be included in the integrated set include but are not limited to: learning time (length of each learning session), practice performance (frequency and accuracy of practice), review status (number and interval of review), and test scores (test scores related to specific knowledge points).

[0040] Construct a knowledge graph, decompose the course content into knowledge points, and then map the integrated set of students' learning trajectory data to the knowledge graph to obtain a learning trajectory-knowledge joint graph; the knowledge graph is a structured representation of knowledge points and their relationships, helping students understand the dependencies between knowledge points. Mapping students' learning trajectory data to the knowledge graph can more clearly see which knowledge points students perform well in and which knowledge points still need to be strengthened. The process of constructing a knowledge graph includes not only the decomposition of knowledge points, but also the associations between knowledge points (such as pre-knowledge and subsequent knowledge). Through this mapping, a graph that combines students' learning behavior and knowledge structure can be generated, called a learning trajectory-knowledge joint graph.

[0041] Based on the learning trajectory-knowledge joint graph, we conduct knowledge point mastery analysis to obtain the knowledge mastery value of each knowledge point, and at the same time conduct knowledge point learning analysis to obtain the knowledge learning rate of each knowledge point; the knowledge mastery value is an indicator to measure the student's mastery of a specific knowledge point, which is calculated based on the student's learning trajectory. The knowledge learning rate indicates the speed or efficiency of students in learning a certain knowledge point, reflecting the student's learning ability for the knowledge point. The calculation of the knowledge mastery value can integrate multiple factors, such as learning time, practice accuracy, review frequency and test scores. The calculation of the knowledge learning rate can be measured by the changes in the knowledge mastery value in different time periods. Through these analyses, we can identify which knowledge points students have mastered well and which still need more time and effort.

[0042] Based on the knowledge mastery value, the mastery degree of knowledge points is judged, and all knowledge points are divided into a set of strong knowledge points and a set of weak knowledge points; by classifying students' knowledge mastery values, students' knowledge points can be divided into strong and weak ones. This can help students learn more targeted, strengthen weak knowledge points, and consolidate strong knowledge points. The classification of knowledge points can be based on the range of knowledge mastery values, and different level standards can be set (for example, mastery values ​​> 0.7 are strong points, and < 0.4 are weak points). This process can not only help students identify their own learning weaknesses, but also provide teachers with feedback on students' learning status.

[0043] For a set of weak knowledge points, the weak recommendation degree is obtained based on the knowledge mastery value and knowledge learning rate, and courses are recommended based on the weak recommendation degree; weak knowledge points need to be processed first to help students make up for the gaps in their learning. The weak recommendation degree combines the knowledge mastery value and the learning rate, and gives priority to recommending courses corresponding to knowledge points with low student mastery and slow learning rates. The weak recommendation degree can be calculated by a formula, giving higher recommendation weights to knowledge points with low knowledge mastery values ​​and slow learning rates. The recommended course content will be more targeted at these weak knowledge points, and course content that highly overlaps with these knowledge points can be given priority. At the same time, recommended courses can help students better master weak knowledge by adjusting the teaching difficulty and providing more practice and review opportunities.

[0044] For a set of strong knowledge points, the knowledge learning rate of the target knowledge point is the strong recommendation degree, and courses are recommended based on the strong recommendation degree; strong knowledge points represent knowledge that students have already mastered well, but still need to consolidate and improve. Recommending strong courses based on the knowledge learning rate can further improve students' learning efficiency and enable them to achieve higher learning results on knowledge points they are good at. The strong recommendation degree is mainly determined based on the students' learning rate. Knowledge points with higher learning rates can recommend more in-depth and complex course content to help students further expand related knowledge or perform advanced applications. Recommended courses may cover extensible knowledge points to help students go further on the basis of their existing strengths.

[0045] Through the above steps, we can not only accurately analyze the students' learning situation, but also recommend courses for students based on their individual weaknesses and strengths. This method improves the pertinence and efficiency of learning, enabling students to better overcome their weaknesses while consolidating and improving their strengths during the learning process.

[0046] Based on the learning trajectory-knowledge joint graph, the knowledge point mastery analysis is performed, and the knowledge mastery value of each knowledge point is obtained, which means:

[0047] Get the total learning time that students spend on target knowledge point i and mark it as , the frequency of students’ review on target knowledge point i and marked as , the correct rate of the student's jth exercise on the target knowledge point i and marked as , the student's test score for the kth test on target knowledge point i and marked as ; Then substitute them into the knowledge mastery value calculation formula of target knowledge point i:

[0048] ; , , are the preset influence coefficients and , , The sum of is one, is a positive value, is the slope parameter of the logistic function, ranging from 0.01 to 10, Review frequency The corresponding preset threshold; n is the total number of exercises, m is the total number of exams, The preset adjustment coefficient for the jth exercise, is the adjustment coefficient for the kth test, is the knowledge mastery value of the target knowledge point i. The larger the value, the higher the students' mastery of the knowledge point. In other words, students have invested more time in learning the knowledge point, have a higher accuracy rate in practice, have a higher review frequency, and have good performance in related exams. A value of 0 means that students have a better learning effect on this knowledge point and have a more solid grasp of it.

[0049] Indicates the correct rate of the student on knowledge point i during the jth exercise. Since students may practice multiple times, the correct rate of each exercise may be different, so it is necessary to summarize or weight the multiple exercises. j represents the exercise number, combined with the preset adjustment coefficient for the jth exercise , weighted processing can be performed to obtain the students’ practice performance on this knowledge point. represents the student's score on knowledge point i in the kth exam. This indicates that multiple exam scores for a specific knowledge point need to be combined to reflect the student's comprehensive performance in different exams. k represents the exam number, combined with the adjustment coefficient for the kth exam , which can comprehensively measure students’ test performance on this knowledge point.

[0050] Total study time The impact of learning is not linear. As time goes by, its marginal benefits will decrease. Therefore, a logarithmic function is used to deal with learning time to avoid overestimating the learning effect after a long time of investment. The logarithmic function can simulate this decreasing effect well.

[0051] The effect of practice accuracy is modified by review frequency. When the review frequency is high, students will have a more significant improvement in the mastery of the practice, so the effect of accuracy increases with the increase of review frequency. However, in order to avoid over-reliance on the effect of review frequency, the square term in the denominator is used to control the effect of review frequency.

[0052] Review frequency affects the performance of practice and test scores, so the effect of review frequency is reflected in the accuracy of practice and test scores. Higher review frequency will theoretically lead to better practice results and better test performance, but this effect will gradually level off as the review frequency increases.

[0053] The impact of test scores is dynamically adjusted by the review frequency. Using the logistic function, when the review frequency is above a certain threshold, When the frequency of review is high, the weight of the test score will increase significantly, and vice versa. This form reflects the "priming" effect of review on test scores, that is, students need to reach a certain review frequency before their test scores will be greatly improved.

[0054] , , Used to balance the contribution of study time, practice accuracy, and test scores to knowledge mastery. Can be adjusted according to the characteristics of the course. A positive value controls the effect of review frequency on the accuracy of practice. The larger the value, the more significant the effect of review frequency. is the slope parameter of the logistic function, which controls the rate at which review frequency improves test scores. The larger the value, the more obvious the effect of review frequency on test scores. The value of is usually positive because it controls the rate and must be positive to ensure the logical rationality of the model. The value of will affect the sensitivity of the model. The larger the value, the faster the reaction; the smaller the value, the slower the reaction. The value range of 0.01 to 10 is a common range. The specific value can also be set according to the size of the data and the change in learning rate.

[0055] It should be noted that in the calculation formula of the knowledge mastery value, the logarithm base can usually be a natural logarithm (base e) or a common logarithm (base 10), depending on the actual application scenario and the desired effect. The natural logarithm (base e) is usually used in scientific computing or mathematical modeling because it is closely related to the exponential growth process, and is particularly suitable for representing the diminishing marginal effect of learning behavior. The natural logarithm is very commonly used in dealing with models such as exponential growth and decay, and is suitable for representing the scenario of diminishing marginal benefits of learning time. Common logarithm (base 10): If you want the result to be more intuitive or more practical in some applications, you can use a logarithm with a base of 10. The common logarithm is usually used for information analysis, which is suitable for use when the data scale is large or the result is expected to be presented in a small numerical range. Due to the diminishing marginal effect of learning behavior, the present invention hopes that the model will more naturally reflect the nonlinear growth or decay in the learning process, and the natural logarithm (base e) can be selected. If in actual use, the user wants the result to be more intuitive and concise within a certain range, the common logarithm (base 10) can be selected.

[0056] Establish a student learning model based on knowledge mastery value: obtain the student's knowledge mastery value at the starting time point of the target knowledge point i and mark it as , the maximum mastery value increment that students can eventually achieve and marked as , build the following model: ; to minimize the predicted value The difference with the actual value is used to determine the attenuation coefficient The numerical value of represents the predicted knowledge mastery value corresponding to time t, and the attenuation coefficient The determined values ​​are substituted into the student learning model for application. By establishing a dynamic change model of knowledge mastery value, the student's knowledge mastery of a certain knowledge point can be estimated based on time. This model not only takes into account the student's initial knowledge mastery level at a certain knowledge point, but also reflects the student's mastery improvement over time through the exponential decay formula, which eventually tends to a certain maximum value. The exponential decay coefficient controls the speed of learning, that is, after a period of learning, the rate of mastery improvement will gradually slow down.

[0057] To analyze the learning of knowledge points and obtain the knowledge learning rate of each knowledge point, we need to obtain the student learning model applied by the students on the target knowledge point i, and then the knowledge learning rate of the students on the target knowledge point i: ; Represents the student's knowledge learning rate at the target knowledge point i. By deriving the formula, the student's learning rate at the target knowledge point is obtained, which describes the student's learning efficiency at different time points. Through this rate, teachers can identify the bottlenecks and progress in students' learning. In the expression of learning rate, the rate is proportional to the student's maximum mastery increment and learning decay parameter, indicating that the initial learning speed is fast, and the learning effect gradually stabilizes over time. This model can provide a basis for personalized learning plans. According to the different learning rates and mastery of students, the recommended learning content can be dynamically adjusted, and the student's learning progress and mastery status can be tracked in real time, helping teachers and students better understand the weak links in the learning process. For educational institutions and teachers, this model helps to identify whether students have low learning efficiency at certain knowledge points during the teaching process, and then provide targeted tutoring and intervention.

[0058] Judging the mastery of knowledge points based on knowledge mastery values ​​means: obtaining the knowledge mastery values ​​of students on each knowledge point and comparing them with the preset knowledge point mastery threshold. If the knowledge mastery value on the knowledge point is greater than or equal to the preset knowledge point mastery threshold, the knowledge point is marked as a learning strength knowledge point. If the knowledge mastery value on the knowledge point is less than the preset knowledge point mastery threshold, the knowledge point is marked as a learning weakness knowledge point. Summarize all the knowledge points corresponding to the learning strength knowledge points to obtain a set of strong knowledge points; summarize all the knowledge points corresponding to the learning weakness knowledge points to obtain a set of weak knowledge points. By classifying the knowledge points into strengths and weaknesses according to the level of knowledge mastery values, it can help the learning system or teachers customize personalized learning paths for students. Strong knowledge points can reduce review, while weak knowledge points need to be focused on and learned. This method can improve learning efficiency and enable students to allocate learning time more effectively. The knowledge mastery value can be continuously updated as time goes by and the learning process progresses. When students’ mastery of weak knowledge points improves, learning strategies can be adjusted dynamically to help students consolidate their strong knowledge points or overcome new weak knowledge points in a timely manner, thereby achieving efficient knowledge point management.

[0059] Obtaining the weak recommendation degree according to the knowledge mastery value and the knowledge learning rate means: mapping the knowledge mastery value and the knowledge learning rate to the numerical range of [0,1] respectively, and obtaining the mapped knowledge mastery value , Knowledge Learning Rate , and then substitute it into the weak recommendation calculation formula:

[0060] ; Recommended for weak points, , Both are preset recommendation coefficients, which respectively measure the impact of knowledge mastery value and learning rate on the recommendation degree of weaknesses, and the sum of the two is one. It is the mapping result of knowledge mastery value, which indicates the degree of students’ mastery of the knowledge point. The larger the value, the better the mastery. The mapping result of knowledge learning rate indicates the learning efficiency of students for this knowledge point. The larger the value, the higher the learning efficiency. The larger the weak recommendation degree, the weaker the knowledge point is, which means that students have not mastered the knowledge point sufficiently or have low learning efficiency. Therefore, relevant courses need to be recommended first.

[0061] Recommending courses based on the degree of weakness means: Arrange in descending order to get the main page list of weak recommendation. Click on each knowledge point on the main page list of weak recommendation to enter the corresponding sub-page list of weak recommendation. The sub-page list of weak recommendation consists of multiple courses containing the knowledge point, and the multiple courses containing the knowledge point are arranged in descending order according to the content similarity between the course and the knowledge point. The purpose of sorting the weak recommendation is to sort the priority recommendation order of the knowledge points according to the previously calculated weak recommendation. The higher the weak recommendation, the weaker the knowledge point, so the corresponding course will be recommended first. Users can click on the knowledge point in the main page of weak recommendation to enter the sub-page list of weak recommendation. Multiple courses related to the knowledge point will be displayed in the sub-page, and they will be sorted according to the content similarity between the course and the knowledge point. In this way, students can see the courses most relevant to their weak knowledge points first and study in a targeted manner.

[0062] For example: suppose a student's weak point is "chain rule in calculus". After analyzing the knowledge mastery value and learning rate, it is found that the weak recommendation of this knowledge point is high. Therefore, "chain rule" is recommended as a key point in the main page of weak recommendation. When the student clicks on this knowledge point, he will enter the sub-page, which displays multiple courses related to the chain rule. These courses may include: Basic course: introduces the definition and basic application of the chain rule. Intermediate course: detailed explanation of the application of the chain rule in different mathematical scenarios. Advanced course: advanced application of the chain rule in multivariable calculus. These courses will be sorted according to the similarity of the content with the chain rule to ensure that the most relevant courses are ranked first, helping students quickly find the most suitable learning resources. Course resources usually come from cloud-based course libraries, which store a large amount of structured learning materials and can be searched by keywords and knowledge points. Cloud-based course libraries can be self-built resources of the education platform or integrated through third-party course providers. These resources will include courses of different difficulty levels, covering content from basic to advanced. The similarity between the course and knowledge point content can be achieved through a variety of existing technologies. Common algorithms include: Similarity calculation based on word vectors (such as TF-IDF or Word2Vec): This method converts the course content and knowledge point description into word vectors and calculates the cosine similarity between them to determine the similarity. The specific similarity calculation process will not be described in detail here.

[0063] For the set of strong knowledge points, the knowledge learning rate of the target knowledge point, i.e., the strong recommendation degree, and the course recommendation based on the strong recommendation degree are executed according to the pre-set strategy, specifically: in the set of strong knowledge points, the knowledge learning rate is an indicator to measure the students' learning efficiency of the knowledge points. The learning rate of strong knowledge points is usually high, indicating that students learn quickly and have a solid grasp of these knowledge points. Therefore, the learning rate can be directly used as the strong recommendation degree. The strong knowledge points are sorted in descending order according to the calculated strong recommendation degree, and the high-level courses corresponding to the knowledge points with the highest learning rate are recommended first. The higher the strong recommendation degree of the knowledge point, the more solid the students' grasp of it is and the faster they learn, so more challenging courses are recommended. There are pre-marked course difficulty labels in the cloud course library or the internal course library of the learning system. By matching the correlation between the strong knowledge points and the knowledge points of the high-level courses, suitable advanced courses can be recommended for students. Assuming that the student has a high knowledge learning rate on the knowledge point of "differential equations", this knowledge point can be listed as a strong point. According to the students' learning situation in "Differential Equations", we can recommend advanced courses, such as "Partial Differential Equations", "Application of Differential Equations in Physics" and other difficult courses to further improve students' abilities in this field. We recommend more advanced and challenging courses to students to ensure that they can further improve in their strong areas. This method can help students expand their knowledge based on their existing advantages and tap into greater learning potential.

[0064] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0065] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0066] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0067] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0068] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A course recommendation method based on learning trajectory and knowledge graph, characterized in that: The following steps are involved: Collect and integrate students' learning trajectory data to obtain an integrated set of students' learning trajectory data; Construct a knowledge graph, decompose the course content into knowledge points, and then map the integrated set of students' learning trajectory data onto the knowledge graph to obtain a learning trajectory-knowledge joint graph; Based on the learning trajectory-knowledge joint graph, the knowledge point mastery analysis is performed to obtain the knowledge mastery value of each knowledge point. At the same time, the knowledge point learning analysis is performed to obtain the knowledge learning rate of each knowledge point. Based on the knowledge mastery value, the mastery degree of knowledge points is judged, and all knowledge points are divided into a strong knowledge point set and a weak knowledge point set; For the weak knowledge point set, the weak recommendation degree is obtained according to the knowledge mastery value and knowledge learning rate, and the course recommendation is made according to the weak recommendation degree; For a set of strong knowledge points, the knowledge learning rate of the target knowledge point is the strong recommendation degree, and courses are recommended based on the strong recommendation degree; Based on the learning trajectory-knowledge joint graph, the knowledge point mastery analysis is performed, and the knowledge mastery value of each knowledge point is obtained, which means: Get the total learning time that students spend on target knowledge point i and mark it as , the frequency of students’ review on target knowledge point i and marked as , the correct rate of the student's jth exercise on the target knowledge point i and marked as , the student's test score for the kth test on target knowledge point i and marked as ; Then substitute them into the knowledge mastery value calculation formula of target knowledge point i: ; , , are the preset influence coefficients and , , The sum of is one, is a positive value, is the slope parameter of the logistic function, ranging from 0.01 to 10, Review frequency The corresponding preset thresholds; n is the total number of exercises, m is the total number of exams, and Tmax represents the maximum learning time of the preset learning knowledge points. The preset adjustment coefficient for the jth exercise, is the adjustment coefficient for the kth test, is the knowledge mastery value of target knowledge point i; Establish student learning model based on knowledge mastery value: Get the student's knowledge mastery value at the starting time point of the target knowledge point i and mark it as , the maximum mastery value increment that students can eventually achieve and marked as , build the following model: ; to minimize the predicted value The difference with the actual value is used to determine the attenuation coefficient The numerical value of represents the predicted knowledge mastery value corresponding to time t, and the attenuation coefficient Substitute the determined values ​​into the student learning model for application; Conduct knowledge point learning analysis and obtain the knowledge learning rate of each knowledge point, which is: Get the student learning model applied by the student on the target knowledge point i, and then the student's knowledge learning rate on the target knowledge point i: ; Represents the student's knowledge learning rate on target knowledge point i.

2. According to claim 1, a course recommendation method based on learning trajectory and knowledge graph is characterized in that: Judging the mastery of knowledge points based on knowledge mastery value means: The student's knowledge mastery value on each knowledge point is obtained respectively and compared with the preset knowledge point mastery threshold. If the knowledge mastery value on the knowledge point is greater than or equal to the preset knowledge point mastery threshold, the knowledge point is marked as a learning strength knowledge point. If the knowledge mastery value on the knowledge point is less than the preset knowledge point mastery threshold, the knowledge point is marked as a learning weakness knowledge point.

3. A course recommendation method based on learning trajectory and knowledge graph according to claim 2, characterized in that: Dividing all knowledge points into a set of strong knowledge points and a set of weak knowledge points means: Summarize all knowledge points marked as learning strengths to obtain a set of strengths; All knowledge points marked as learning weaknesses are summarized to obtain a set of weak knowledge points.

4. A course recommendation method based on learning trajectory and knowledge graph according to claim 3, characterized in that: The recommended degree of weakness based on knowledge mastery value and knowledge learning rate refers to: Map the knowledge mastery value and knowledge learning rate to the numerical range of [0,1] respectively to obtain the mapped knowledge mastery value , Knowledge Learning Rate , and then substitute it into the weak recommendation calculation formula: ; Recommended for weak points, , Both are preset recommendation coefficients and their sum is one.

5. According to claim 4, a course recommendation method based on learning trajectory and knowledge graph is characterized in that: Recommending courses based on weakness means: Recommend weak points Arrange in descending order to obtain the weakness recommendation main page list. Click on each knowledge point on the weakness recommendation main page list to enter the weakness recommendation sub-page list corresponding to the knowledge point. The weakness recommendation sub-page list is composed of multiple courses containing the knowledge point, and the multiple courses containing the knowledge point are arranged in descending order according to the content similarity between the course and the knowledge point.

Images