A course recommendation method and system considering time characteristics and multiple learning motivations

Abstract

The application discloses a course recommendation method and system considering time characteristics and various learning motivations, comprising: modeling student, course and time characteristic embedding; modeling student course selection motivation; recommending courses selected by students in the future. The application uses technical methods such as time sequence point models, attention mechanisms and self-determination theory, systematically analyzes historical learning record data of students, calculates different contributions of different historical interaction records to the recommendation result by using time characteristics in the interaction, and considers four course selection motivations, so that the accuracy and interpretability of course recommendation are improved.

Description

Technical Field

[0001] This invention relates to the field of big data and intelligent technology, and in particular to a course recommendation method and system that takes into account time characteristics and multiple learning motivations. Background Technology

[0002] In recent years, course recommendation algorithms, as a crucial technology for filtering educational information, have attracted considerable attention from researchers. Numerous models have been developed for course recommendation algorithms, including those based on collaborative filtering, graph neural networks, and reinforcement learning. Despite the good results achieved by current course recommendation algorithms, they still face the following two challenges: (1) These methods only utilize static student-course interactions as input, assuming that the only factors influencing the next recommended course are previously learned courses and their order. However, student preferences change dynamically with time and historical interactions, and the temporal characteristics of these interactions reflect important information such as students' learning cycles and course scheduling. For example, in Figure 1 In this scenario, both student A and student C have studied advanced mathematics. Based on past experience, the recommendation system might recommend linear algebra at time t7. It's worth noting that student A studied advanced mathematics at time t6, closer to the recommendation time t7. Student C studied advanced mathematics at time t1, further back from t7. Even though both students studied advanced mathematics, the difference in their interaction times will have different impacts on recommending linear algebra at time t7. Previous course recommendation algorithms failed to capture these differences.

[0003] (2) In recent model designs, most researchers have opted for complex models to achieve better accuracy at the expense of interpretability. Interpretability is a crucial indicator for improving the persuasiveness, effectiveness, and user satisfaction of recommendation systems. Course recommendation algorithms should not only recommend suitable courses but also clearly explain the students' learning motivations for choosing those courses. This latent information is embedded in the students' historical interaction sequences. For example… Figure 1 In the diagram, the red arrows represent influence. Student A's choice influenced Student B's subsequent choice, possibly because Student B saw Student A as a role model and followed her decision. Similarly, at time t4, both Student C and Student A chose Data Structures, demonstrating the influence of conformity. Furthermore, at time t7, Chatgpt was very popular, increasing the probability of the system recommending this course, reflecting the influence of popular courses. Previous course recommendation algorithms did not comprehensively examine the factors influencing student course selection, thus failing to provide a convincing explanation. Summary of the Invention

[0004] The main objective of this invention is to provide a knowledge tracking method and system based on answer status and multidimensional item response theory to solve the above-mentioned technical problems.

[0005] To achieve the above objectives, this invention provides a course recommendation method that considers time characteristics and multiple learning motivations, comprising the following steps: S1 models the embeddings of student, course, and time features, including: S11 uses two different embedding layers to map student IDs and course IDs to a low-dimensional latent space, where let U∈R |m|*d e represents the user embedding matrix generated by the user embedding layer. uj Let m be the embedding vector of student j, where m is the number of students and d is the embedding size; let C ∈ R |n|*d e represents the user embedding matrix generated by the user embedding layer. ci Let be the embedding vector of course i, where n is the number of students and d is the embedding size; the student and course vectors are randomly initialized and iteratively updated along with the loss function. S12, Determine the time vector representation based on time characteristics. ; S2, Learning Motivation Modeling, constructing a time-series model based on learning motivation; S3, Recommendation Result Prediction: Input students' course selection motivation into the time series model to make predictions and obtain the probability of recommendation.

[0006] Preferably, step S12 includes: S121, Absolute Time Embedding, the absolute time vector representation of timestamp tL is as follows: in From the monthly embedding matrix M m The monthly vector of the timestamp tL extracted from it. From the daily embedding matrix M d The daily vector of the timestamp tL extracted from it. Let be the absolute time vector of timestamp tL; the month and day embeddings are randomly initialized and iteratively updated along with the loss function; S122, relative time embedding, let: ; ; Where pos represents the position of the word, i represents the dimension, and d represents the position of the word. model Indicates the dimensionality of the model, even-numbered digits and odd number When concatenated together, they represent the relative time vector of timestamp tL. The relative time vector is fixed and does not update iteratively along with the loss function; S123, concatenate the absolute time vector and the relative time vector together to obtain the final time vector representation. : Preferably, S2 includes: S21, Basic Strength Modeling: The d-dimensional embeddings of students and courses are input into the strength function, as shown in the formula: ; in, This represents the probability of recommending course CI to student UJ without considering any historical interactions; S22, Intrinsic learning motivation modeling, based on students' historical course selection records. Calculate the attention weight of each history course to the candidate courses. : ;in, , It is embedded in the student's UJ history course. It is a trainable weight matrix. , This indicates the rate of decline of historical influence. It is an embedding of the recommendation time. It's a student's UJ. Interactive time embedding of a history lesson; the value of intrinsic learning motivation for: ,in, ; S23, External learning motivation modeling, including: Students whose attention relationships are aggregated using attention weights The information is used to obtain a score for the relationship of interest. According to the formula: in, This represents the attention weight of each attention relationship to the candidate courses. W1 is the embedding of the student's attention relationships, and W2 is a trainable weight matrix. , This indicates the rate of decline of historical influence. It is an embedding of recommendation time. It's a student's UJ. If an embedding that focuses on student interaction time is used, then... The student who most recently interacted with the student at recommendation time t is selected as ucj, and a score for conformity is obtained. According to the formula ,formula as well as get ; in, The recommendation moment is when students are encouraged to conform and become embedded in the student's mindset. It is a trainable weight matrix; Filter the courses that have generated the most recent interaction up to recommendation time t, and use an attention mechanism to determine the influence weight of each popular course. This leads to the score for popular courses. According to the formula ,formula and formula get ;in, Embedding of popular courses It is a trainable weight matrix.

[0007] Preferably, step S22 includes the step where all users uniformly select the most recent A number of interaction records, if the length exceeds Remove redundant interactions; if the length is less than the specified length, fill with a zero vector; obtain the student's historical course selection record. .

[0008] Preferably, step S3 includes: Based on the values ​​of basic intensity, intrinsic learning motivation, and extrinsic learning motivation calculated in step S2, substitute them into the formula: To obtain the recommended strength ; Normalize the results using sigmoid to obtain the final recommendation probability. .

[0009] Preferably, binary cross-entropy is used as the loss function, and the loss function is defined for the training set as: represents a positive example, For negative examples, the purpose of the training process is to optimize the model parameters so that the outputs of uj and ci at time t of each interaction are close to 1, and the outputs of uj and ci′ are as close to 0 as possible; all parameters and embeddings are learned in an end-to-end training manner through backpropagation.

[0010] The present invention also provides a course recommendation system that considers time characteristics and multiple learning motivations, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the course recommendation method that considers time characteristics and multiple learning motivations as described in any of the preceding claims.

[0011] The present invention provides a course recommendation method and system that considers time features and multiple learning motivations, including: (1) modeling student, course, and time feature embeddings; (2) modeling student course selection motivations; and (3) recommending courses for students to choose in the future. The present invention utilizes techniques such as time-series models, attention mechanisms, and self-determination theory to systematically analyze students' historical learning record data, calculates the different contributions of different historical interaction records to the recommendation results using time features in the interactions, and considers four course selection motivations to improve the accuracy and interpretability of course recommendations. Attached Figure Description

[0012] The accompanying drawings, as part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments and descriptions of the invention are used to explain the invention, but do not constitute an undue limitation of the invention. Obviously, the drawings described below are merely some embodiments, and those skilled in the art can obtain other drawings based on these drawings without creative effort. In the drawings: Figure 1 This is a schematic diagram illustrating student course selection sequences under different circumstances.

[0013] Figure 2 This is a schematic diagram of the process for obtaining a student's knowledge status in one embodiment of the present invention.

[0014] Figure 3 This is a flowchart illustrating the time embedding step in one embodiment of the present invention.

[0015] Figure 4 This is a flowchart illustrating a course recommendation method that considers time characteristics and multiple learning motivations in one embodiment of the present invention.

[0016] Figure 5 This is a schematic diagram of the hardware structure for running a course recommendation method based on considering time characteristics and multiple learning motivations in one embodiment of the present invention.

[0017] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] The technical problems solved by the embodiments of the present invention, the technical solutions adopted, and the technical effects achieved will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other equivalent or obvious variations of embodiments obtained by those skilled in the art without creative effort fall within the protection scope of the present invention. The embodiments of the present invention can be embodied in various different ways as defined and covered by the claims.

[0019] It should be noted that many specific details are given in the following description for ease of understanding. However, it is obvious that the present invention may be implemented without these specific details.

[0020] It should be noted that, in the absence of explicit limitations or conflicts, the various embodiments and their technical features in this invention can be combined with each other to form a technical solution.

[0021] This invention proposes a course recommendation method that considers time characteristics and multiple learning motivations, including the following steps: S1, modeling embeddings of student, course, and time features; S2, Learning Motivation Modeling, constructing a time-series model based on learning motivation; S3, Recommendation Result Prediction: Input students' course selection motivation into the time series model to make predictions and obtain the probability of recommendation.

[0022] Specifically, S1 includes: S11. Due to the sparsity and high dimensionality of student and course IDs, transforming them to a low-dimensional space is a typical practice. Usually, two different embedding layers are used to map student and course IDs to the low-dimensional latent space. Two different embedding layers are used to map student IDs and course IDs to the low-dimensional latent space, where let U∈R |m|*d e represents the user embedding matrix generated by the user embedding layer. uj Let m be the embedding vector of student j, where m is the number of students and d is the embedding size; let C ∈ R |n|*d e represents the user embedding matrix generated by the user embedding layer. ci Let be the embedding vector of course i, where n is the number of students and d is the embedding size; the student and course vectors are randomly initialized and iteratively updated along with the loss function. S12, Determine the time vector representation based on time characteristics. .

[0023] Preferably, step S12 includes: S121, Absolute Time Embedding, the absolute time vector representation of timestamp tL is as follows: ;in, From the monthly embedding matrix M m The monthly vector of the timestamp tL extracted from it. From the daily embedding matrix M d The daily vector of the timestamp tL extracted from it. Let be the absolute time vector of timestamp tL; the month and day embeddings are randomly initialized and iteratively updated along with the loss function; S122, relative time embedding, let: , ; Where pos represents the position of the word, i represents the dimension, and d represents the position of the word. model Indicates the dimensionality of the model, even-numbered digits and odd number When concatenated together, they represent the relative time vector of timestamp tL. The relative time vector is fixed and does not update iteratively along with the loss function; S123, concatenate the absolute time vector and the relative time vector together to obtain the final time vector representation. : .

[0024] In step S121, to generate the timestamp embedding, time information needs to be considered at multiple granularities to enrich the time representation. In this paper, the granularity of month and day is considered. The absolute time vector representation of timestamp tL is as follows: .

[0025] In step S122, a position vector is added to capture the relative order of items in the student or course domain of the sequence. This paper adopts the position encoding method proposed by Vaswani et al., which does not introduce additional parameters and can be extrapolated to sequence lengths longer than a predefined fixed length.

[0026] Preferably, S2 includes: S21, Basic strength modeling; S22, Modeling intrinsic learning motivation; S23, Modeling of extrinsic learning motivation.

[0027] Specifically, in step S21, regardless of what courses a student has previously taken, all courses have a certain probability of being recommended to the student; this is called baseline strength. Baseline strength modeling includes inputting the d-dimensional embeddings of the student and courses into the strength function, as shown in the formula: ; in, This represents the probability of recommending course CI to student UJ without considering any historical interactions.

[0028] Specifically, in step S22, after obtaining the interaction records for each student, the interest preference values ​​for the candidate courses ci are calculated. Specifically, the attention weight of each historical course to the candidate courses must first be calculated. .

[0029] Attention weights are derived from two parts: the course vector and the time vector; secondly, they are derived from the formula... By aggregating historical course information that students have interacted with on UJ, a final score based on their interest preferences is obtained.

[0030] Modeling intrinsic learning motivation includes: based on students' historical course selection records Calculate the attention weight of each history course to the candidate courses. , ; in, , It is embedded in the student's UJ history course. It is a trainable weight matrix. , This indicates the rate of decline of historical influence. It is an embedding of the recommendation time. It's a student's UJ. Interactive time embedding of a history lesson; the value of intrinsic learning motivation for: ,in, .

[0031] Preferably, step S22 includes the step where all users uniformly select the most recent A number of interaction records, if the length exceeds Remove redundant interactions; if the length is less than the specified length, fill with a zero vector; obtain the student's historical course selection record. .

[0032] This invention obtains students' interests and preferences from their historical records. For ease of subsequent calculations, all users are uniformly selected based on their most recent... One interaction record. If the length exceeds... Remove redundant interactions; if the length is less than the specified length, fill with a zero vector.

[0033] Specifically, in step S23, regarding the modeling of extrinsic learning motivation, this paper considers students' attentional relationships, conformity, and the influence of popular courses. This motivation often stems from the external environment and plays an indispensable role in recommendations. These will be discussed in detail below.

[0034] Students are not independent of each other; they often have peer relationships. When choosing courses, students tend to gravitate towards high-achieving or close classmates. Therefore, this paper will analyze the interaction sequence S of student uj. uj Compare with the interaction sequences of other students. If student uv has also taken a course that student uj has taken, and took that course before student uj, then student uv is one of student uj's attention relationships. Similarly, a fixed length is chosen here. Similar to interest preferences, attention weights are used to aggregate information about students' attention relationships among all students' attention relationships, and a score for attention relationship is obtained.

[0035] Students whose attention relationships are aggregated using attention weights The information is used to obtain a score for the relationship of interest. According to the formula: ; in, This represents the attention weight of each attention relationship to the candidate courses. , W1 is the embedding of the student's attention relationships, and W2 is a trainable weight matrix. , This indicates the rate of decline of historical influence. It is an embedding of the recommendation time. It's a student's UJ. If an embedding that focuses on student interaction time is used, then... Select the students who have most recently interacted with you at the recommendation time t. To obtain a score for herd mentality According to the formula ,formula and formula get ; in, The recommendation moment is when students are encouraged to conform and become embedded in the student's mindset. It is a trainable weight matrix. Filter the courses that have generated the most recent interaction up to recommendation time t, and use an attention mechanism to determine the influence weight of each popular course. This leads to the score for popular courses. According to the formula ,formula and formula get ; in, Embedding of popular courses It is a trainable weight matrix.

[0036] Preferably, step S3 includes: Based on the values ​​of basic intensity, intrinsic learning motivation, and extrinsic learning motivation calculated in step S2, substitute them into the formula: To obtain the recommended strength ; Normalize the results using sigmoid to obtain the final recommendation probability. .

[0037] in, .

[0038] Preferably, binary cross-entropy is used as the loss function, and the loss function is defined for the training set as: represents a positive example, For negative examples, the purpose of the training process is to optimize the model parameters so that the outputs of uj and ci at time t of each interaction are close to 1, and the outputs of uj and ci′ are as close to 0 as possible; all parameters and embeddings are learned in an end-to-end training manner through backpropagation.

[0039] The present invention provides a course recommendation method and system that considers time features and multiple learning motivations, including: (1) modeling student, course, and time feature embeddings; (2) modeling student course selection motivations; and (3) recommending courses for students to choose in the future. The present invention utilizes techniques such as time-series models, attention mechanisms, and self-determination theory to systematically analyze students' historical learning record data, calculates the different contributions of different historical interaction records to the recommendation results using time features in the interactions, and considers four course selection motivations to improve the accuracy and interpretability of course recommendations.

[0040] The present invention also provides a course recommendation system that considers time characteristics and multiple learning motivations, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the course recommendation method that considers time characteristics and multiple learning motivations as described in any of the preceding claims.

[0041] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the course recommendation method considering time characteristics and multiple learning motivations as described in any of the preceding claims.

[0042] Figure 5 This is a schematic diagram of the hardware structure for running a course recommendation method that considers time characteristics and multiple learning motivations, according to an embodiment of the present invention. Figure 3 As shown, this embodiment / computer 6 includes: a processor 60, a memory 61, and a computer program 62 stored in the memory 61 and executable on the processor 60, such as a program for running a course recommendation method that considers time characteristics and multiple learning motivations. When the processor 60 executes the computer program 62, it implements the steps in the various embodiments of running a course recommendation method that considers time characteristics and multiple learning motivations. Alternatively, when the processor 60 executes the computer program 62, it implements the functions of each module / unit in the various device embodiments described above.

[0043] For example, the computer program 62 may be divided into one or more modules / units, which are stored in the memory 61 and executed by the processor 60 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 62 in the computer 6.

[0044] The computer 6 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. The computer 6 may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art will understand that... Figure 3 This is merely an example of computer 6 and does not constitute a limitation on computer 6. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer 6 may also include input / output devices, network access devices, buses, etc.

[0045] The processor 60 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0046] The memory 61 can be an internal storage unit of the computer 6, such as a hard drive or memory. The memory 61 can also be an external storage device of the terminal device 6, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device. Furthermore, the memory 61 can include both internal storage units and external storage devices of the computer 6. The memory 61 is used to store the computer program and other programs and data required by the terminal device. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0047] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0048] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0049] Those skilled in the art will recognize that the units and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0050] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0051] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0052] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0053] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0054] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A course recommendation method that considers time characteristics and multiple learning motivations, characterized in that, Including the following steps: S1 models the embeddings of student, course, and time features, including: S11 uses two different embedding layers to map student IDs and course IDs to a low-dimensional latent space, where let U∈R |m|*d e represents the user embedding matrix generated by the user embedding layer. uj Let m be the embedding vector of student j, where m is the number of students and d is the embedding size; let C ∈ R |n|*d e represents the user embedding matrix generated by the user embedding layer. ci Let be the embedding vector of course i, where n is the number of students and d is the embedding size; the student and course vectors are randomly initialized and iteratively updated along with the loss function. S12, Determine the time vector representation based on time characteristics. ; S2, learning motivation modeling, constructing a time-series model based on learning motivation; S2 includes: S21, Basic Strength Modeling: The d-dimensional embeddings of students and courses are input into the strength function, as shown in the formula: ； in, This represents the probability of recommending course CI to student UJ without considering any historical interactions; The vector representation of student uj, A vector representation of course ci; S22, Intrinsic learning motivation modeling, based on students' historical course selection records. Calculate the attention weight of each history course to the candidate courses. : ;in, ， It is embedded in the student's UJ history course. It is a trainable weight matrix. , This indicates the rate of decline of historical influence. It is an embedding of the recommendation time. It's a student's UJ. Interactive time embedding of a history lesson; the value of intrinsic learning motivation for: ,in, ; express The attention weight of the r-th record in the dataset for the candidate courses. express The rth record in the document is embedded in the UJ history course for middle school students; S23, External learning motivation modeling, including: Students whose attention relationships are aggregated using attention weights The information is used to obtain a score for the relationship of interest. According to the formula: ； ； in, This represents the attention weight of each attention relationship to the candidate courses. W1 is the embedding of the student's attention relationships, and W2 is a trainable weight matrix. , This indicates the rate of decline of historical influence. It is an embedding of the recommendation time. It's a student's UJ. If an embedding that focuses on student interaction time is used, then... Select the students who have most recently interacted with you at the recommendation time t. To obtain a score for herd mentality According to the formula ,formula as well as get ; in, This represents the attention weight of each student who exerts a herd influence on the current student. The recommendation moment is when students are encouraged to conform and become embedded in the student's mindset. It is a trainable weight matrix; Filter the courses that have generated the most recent interaction up to recommendation time t, and use an attention mechanism to determine the influence weight of each popular course. This leads to the score for popular courses. According to the formula ,formula and formula get ;in, Embedding of popular courses It is a trainable weight matrix; S3, Recommendation Result Prediction: Input students' course selection motivation into the time series model to make predictions and obtain the probability of recommendation.

2. The course recommendation method considering time characteristics and multiple learning motivations according to claim 1, characterized in that, Step S12 includes: S121, Absolute Time Embedding, the absolute time vector representation of timestamp tL is as follows: ;in, From the monthly embedding matrix M m The monthly vector of the timestamp tL extracted from it. From the daily embedding matrix M d The daily vector of the timestamp tL extracted from it. Let be the absolute time vector of timestamp tL; the month and day embeddings are randomly initialized and iteratively updated along with the loss function; S122, relative time embedding, let: ； ； Where pos represents the position of the word, i represents the dimension, and d represents the position of the word. model Indicates the dimensionality of the model, even-numbered digits and odd number When concatenated together, they represent the relative time vector of timestamp tL. The relative time vector is fixed and does not update iteratively along with the loss function; S123, concatenate the absolute time vector and the relative time vector together to obtain the final time vector representation. : .

3. The course recommendation method considering time characteristics and multiple learning motivations according to claim 1, characterized in that, Step S22 includes the step where all users uniformly select the most recent A number of interaction records, if the length exceeds Remove unnecessary interactions; If the length is less than the specified length, fill with a zero vector; obtain the student's historical course selection record. .

4. The course recommendation method considering time characteristics and multiple learning motivations according to claim 3, characterized in that, Step S3 includes: Based on the values ​​of basic intensity, intrinsic learning motivation, and extrinsic learning motivation calculated in step S2, substitute them into the formula: To obtain the recommended strength ; Normalize the results using sigmoid to obtain the final recommendation probability. .

5. The course recommendation method considering time characteristics and multiple learning motivations according to any one of claims 1-4, characterized in that, Using binary cross-entropy as the loss function, the loss function for the training set is defined as: represents a positive example, For negative examples, the purpose of the training process is to optimize the model parameters so that the outputs of uj and ci at time t of each interaction are close to 1, and the outputs of uj and ci′ are as close to 0 as possible; all parameters and embeddings are learned in an end-to-end training manner through backpropagation.

6. A course recommendation system considering time characteristics and multiple learning motivations, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the course recommendation method that takes into account time characteristics and multiple learning motivations as described in any one of claims 1 to 5.

Images