CN113722591B - Online course recommendation method - Google Patents
Online course recommendation method Download PDFInfo
- Publication number
- CN113722591B CN113722591B CN202111011484.6A CN202111011484A CN113722591B CN 113722591 B CN113722591 B CN 113722591B CN 202111011484 A CN202111011484 A CN 202111011484A CN 113722591 B CN113722591 B CN 113722591B
- Authority
- CN
- China
- Prior art keywords
- vector
- course
- students
- courses
- student
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 239000013598 vector Substances 0.000 claims abstract description 220
- 238000012512 characterization method Methods 0.000 claims abstract description 28
- 238000013507 mapping Methods 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 12
- 239000000203 mixture Substances 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 230000001186 cumulative effect Effects 0.000 claims 1
- 238000010276 construction Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011423 initialization method Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
- G06Q50/205—Education administration or guidance
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Educational Technology (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an online course recommendation method, which comprises the following steps: taking courses, students, schools and teachers as nodes in advance, and establishing an entity relation model of the courses about the students, schools and teachers respectively; acquiring course association node information associated with a target student of a course to be recommended according to the established entity relation model; calculating multi-hop information propagation vectors of the target students and each course association node, and taking the multi-hop information propagation vectors as characterization vectors of the target students and the course association nodes; according to the characterization vector of the target student and the characterization vector of each course association node, the recommended course of the target student is completed; the system can comprehensively consider courses recommended to students by using the entity relations between the courses and the students, the schools and the teachers respectively, can comprehensively use the relations among multiple entities, improves the accuracy of course recommendation, and simultaneously, after each round of recommendation, the system updates the recommendation model by using the feedback of the students, thereby ensuring the real-time performance of recommendation.
Description
Technical Field
The invention relates to an online course recommendation method, and belongs to the technical field of recommendation systems.
Background
With the development of the Internet, online education is vigorous, a large number of students, teachers and schools participate in the online education, courses of an online education platform are very rich, a learner can select more, a large amount of time is wasted on searching and selecting courses, the learning efficiency is reduced, the learning quality is influenced, and the initiative of the learner is reduced. The course recommendation system can recommend proper courses for students, reduce the time wasted by the students in selecting courses, and enable the learner to concentrate on course learning instead of course selection. In the traditional course online recommendation method, inherent characteristics of courses are used for representing the courses, online recommendation is carried out on the courses, relationships between the courses and the entities such as teachers, schools and students are not well utilized for recommendation, the utilization rate of recommended resources is low, and the accuracy of a recommendation system is low.
The existing course online recommendation method utilizing various entity relations only utilizes the relations among courses, teachers and schools, does not consider the relation of the students in selecting courses, and is an important reason for low accuracy of a course online recommendation system.
The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person of ordinary skill in the art.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides an online course recommendation method and system, which are used for solving the technical problem that the accuracy of a course recommendation system is low because courses are recommended for students only by using the relation between courses and teachers and schools in the prior art.
In order to solve the technical problems, the invention is realized by adopting the following technical scheme:
an online course recommendation method, the recommendation method comprising:
taking courses, students, schools and teachers as nodes in advance, and respectively establishing entity relation models of the courses, the students, the schools and the teachers;
acquiring course association node information associated with a target student of a course to be recommended according to the established entity relation model;
calculating multi-hop information propagation vectors of the target students and each course association node, and taking the multi-hop information propagation vectors as characterization vectors of the target students and the course association nodes;
and recommending courses to the target students according to the characterization vectors of the target students and the characterization vectors of the course association nodes, and completing the course recommendation to the target students.
As a preferable technical scheme of the invention, the step of recommending courses to the target students further comprises the following steps:
acquiring feedback information of students on recommended courses, and calculating rewarding mapping vectors of the courses on the students according to the feedback information;
and updating the course recommended for the student according to the bonus mapping vector of the course to the student.
As a preferable technical scheme of the invention, the calculation formula for updating the rewards mapping vector is as follows:
wherein: a is that l For updating the intermediate amount of the bonus map vector in the first round, T is the vector transpose of Z,a mixed characterization vector for the student and the recommended course in the first round;
wherein b l To update the intermediate quantity of the bonus map vector, r l Feedback for the first round of students on recommended courses,representing vectors for a mixture of students and courses in the first round;
wherein:representing vectors for a mix of students and courses in round I,>for the token vector of the first round of students,representing a vector for a first course, wherein T is the transposition of the vector;
wherein: θ l For a round of bonus mapping vectors, A l More in the first roundIntermediate quantity of new bonus map vector, b l The intermediate amount of the bonus map vector is updated for the first round.
As a preferable technical scheme of the invention, when students select recommended courses, r l =1, when student does not select recommended course, r l ＝0。
As a preferred technical scheme of the invention, the course recommendation to the target students comprises the following steps:
calculating courses recommended to students for the first round based on the following formula;
wherein c l For the first round of courses recommended to students, C is a set of courses, θ is a bonus mapping vector, T is a vector transpose of Z,for a mixed characterization vector of students and courses, alpha is an exploration factor, and A is an intermediate quantity for updating a reward mapping vector;
and sending the calculated courses recommended to the students for the first round to the target students.
As a preferable technical scheme of the invention, the 0 th hop propagation vector of each node in the entity relation model is corresponding to the initial vector of each node in the entity relation model;
the characteristic vectors of the three entities of the course, the teacher and the school are respectively corresponding to an initial vector of the course, an initial vector of the teacher and an initial vector of the school;
the initial vector of the student is an average of the initial vectors of all courses selected by the student.
As a preferable technical scheme of the invention, the calculation method of the multi-hop information propagation vector of the target student comprises the following steps:
acquiring last multi-hop information propagation vectors of all course association nodes associated with the target student node;
calculating the average value of the k-1 jump information propagation vectors of all course associated nodes to obtain the k-1 jump information propagation vector of the course associated node;
and acquiring a kth-1-hop information propagation vector of the target student node, and inputting the kth-1-hop information propagation vector of the target student node and the last-time multi-hop information propagation vector of the course-associated node into a neural network to obtain the kth-hop information propagation vector of the target student node.
As a preferable technical scheme of the invention, the calculation formula of the multi-hop information propagation vector of the target student node is as follows:
wherein,propagation vector of kth hop information of target student node, W k And B k Are trainable parameter matrices, < >>Propagation vector for kth-1 hop of course-associated node,>a propagation vector of the kth-1 hop of the target student node a;
wherein,the kth-1 hop propagation vector of the course-associated node, N (a) is the course-associated node,>the propagation vector of the kth-1 hop for the jth course-associated node.
As a preferable technical scheme of the invention, the method further comprises the step of updating the system parameters, and the method for updating the system parameters comprises the following steps:
acquiring a feedback value of a student actually on a recommended course in the last accumulated error back propagation and a feedback value of an expected student on the recommended course;
calculating accumulated errors of the obtained actual feedback value and the expected feedback value;
counter-propagating the accumulated errors, and completing updating of system parameters by a random gradient descent method;
wherein the system parameters include a trainable parameter matrix, an initial vector of a course, an initial vector of a teacher, and an initial vector of a school.
As a preferred technical solution of the present invention, the inverse error propagation is completed through a mean square loss function, and the calculation formula of the inverse error propagation is:
wherein H is the system accumulated error, θ is the rewarding mapping vector, z is the hybrid characterization vector, L is the system recommended number between two accumulated error counter-propagation, r l Feedback for the first round of students on recommended courses,for the mixed characterization vector of students and courses in the first round, theta l-1 A bonus mapping vector for round l-1;
the update formula for the trainable parameter matrix is:
wherein W is k As a trainable parameter matrix, beta 1 For learning rate, H is the accumulated error of the system;
wherein B is k As a trainable parameter matrix, beta 2 For learning rate, H is the accumulated error of the system;
the update formula for the course initial vector is:
wherein v is c Beta, the initial vector of course 3 For learning rate, H is the accumulated error of the system;
the updating formula of the initial vector of the teacher is as follows:
wherein v is t As initial vector of teacher, beta 4 For learning rate, H is the accumulated error of the system;
the update formula for the initial vector of the school is as follows:
wherein v is u As initial vector of school, beta 4 For learning rate, H is the system accumulated error.
Compared with the prior art, the invention has the beneficial effects that:
according to the invention, the physical relations between courses and students, schools and teachers can be utilized to comprehensively consider the courses recommended to the students, the relations among multiple entities can be comprehensively utilized to further improve the accuracy of course recommendation, and meanwhile, after each round of recommendation, the system updates the recommendation model by utilizing the feedback of the students, so that the real-time performance of recommendation is ensured. After L rounds of continuous recommendation, the parameters and variables in the system are optimized by utilizing accumulated errors, so that the accuracy of recommendation is ensured. The system can fully utilize the recommended resources, and simultaneously ensure the real-time performance and accuracy of recommendation.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of the present invention;
FIG. 3 is a model of the physical relationship of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are merely for convenience in describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first", "a second", etc. may explicitly or implicitly include one or more such feature. In the description of the present invention, unless otherwise indicated, the meaning of "a plurality" is two or more.
In the description of the present invention, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art in a specific case.
As shown in fig. 1 to 3, an online course recommendation method, the recommendation method includes:
taking courses, students, schools and teachers as nodes in advance, and respectively establishing entity relation models of the courses, the students, the schools and the teachers;
acquiring course association node information associated with a target student of a course to be recommended according to the established entity relation model;
calculating multi-hop information propagation vectors of the target students and each course association node, and taking the multi-hop information propagation vectors as characterization vectors of the target students and the course association nodes;
and recommending courses to the target students according to the characterization vectors of the target students and the characterization vectors of the course association nodes, and completing the course recommendation to the target students.
At the beginning of system operation, each parameter in the system needs to be initialized, and the characteristic vectors of three entities of a course, a teacher and a school are respectively corresponding to the initial vector of the course, the initial vector of the teacher and the initial vector of the school, wherein the characteristic vectors and the initial vectors of the entities are D-dimension.
The initial vector of the student is the average of the initial vectors of all courses selected by the student.
Vector representation of students using all courses selected by students, for selected courses { c } 1 ,…,c k ,…,c N Students with initial vector of } beingThe initial vector dimension of the student is D-dimension.
The steps of recommending courses to the target students further comprise:
acquiring feedback information of students on recommended courses, and calculating rewarding mapping vectors of the courses on the students according to the feedback information;
and updating the course recommended for the student according to the bonus mapping vector of the course to the student.
When the system recommends courses actually given to students, the system comprehensively considers the rewarding mapping vector, the characterization vector of the target students and the characterization vector of the courses, and accurately recommends courses for the students.
The system initializes the reward mapping vector theta in the recommendation model, and the initialization method comprises the following steps:
θ 0 ＝A 0 -1 b 0 ；
wherein the method comprises the steps ofAnd the student course interaction history is set to be empty.
A 0 To update the intermediate quantity of the bonus map vector b 0 To update the intermediate amount of the bonus map vector.
The calculation formula for updating the bonus mapping vector is:
wherein: a is that l For updating the intermediate amount of the bonus map vector in the first round, T is the vector transpose of Z,a mixed characterization vector for the student and the recommended course in the first round;
wherein b l To update the intermediate quantity of the bonus map vector, r l Feedback for the first round of students on recommended courses,representing vectors for a mixture of students and courses in the first round;
wherein the method comprises the steps ofRepresenting vectors for a mix of students and courses in round I,>for the token vector of the first round of students,representing a vector for a first course, wherein T is the transposition of the vector;
θ l ＝A l -1 b l ；
wherein: θ l For a round of bonus mapping vectors, A l Updating the intermediate quantity of the bonus map vector, b, for the first round l The intermediate amount of the bonus map vector is updated for the first round.
The accuracy of recommending courses for students is gradually increased by updating the rewarding mapping vector in the continuous recommending process.
Recommending courses to a target student includes the steps of:
calculating courses recommended to students for the first round based on the following formula;
wherein c l For the first round of courses recommended to students, C is a set of courses, θ is a bonus mapping vector, T is a vector transpose of Z,for a mixed characterization vector of students and courses, alpha is an exploration factor, and A is an intermediate quantity for updating a reward mapping vector;
and sending the calculated courses recommended to the students for the first round to the target students.
When the student selects the recommended course, r l =1, when student does not select recommended course, r l ＝0。
Namely, the feedback information of the student selection course is collected, and the updating of the rewarding and rewarding mapping vector is completed through the updating method of the rewarding mapping vector theta.
The 0 th hop propagation vector of each node in the entity relation model is corresponding to the initial vector of each node in the entity relation model, wherein the initial vector value is given to each node in the initial operation process of the system, namely the first round of recommendation to students is completed based on the initial vector value, and the system is gradually perfected based on the subsequent recommendation process.
The 0 th hop propagation vector of each node in the entity relation model corresponds to the initial vector of each node in the entity relation model;
the characteristic vectors of the three entities of the course, the teacher and the school are respectively corresponding to the initial vector of the course, the initial vector of the teacher and the initial vector of the school;
the initial vector of the student is the average of the initial vectors of all courses selected by the student.
The calculation method of the multi-hop information propagation vector of the target student comprises the following steps:
acquiring last multi-hop information propagation vectors of all course association nodes associated with the target student node;
calculating the average value of the k-1 jump information propagation vectors of all course associated nodes to obtain the k-1 jump information propagation vector of the course associated node;
and acquiring a kth-1-hop information propagation vector of the target student node, and inputting the kth-1-hop information propagation vector of the target student node and the last-time multi-hop information propagation vector of the course-associated node into a neural network to obtain the kth-hop information propagation vector of the target student node.
Taking the target teacher node as an example:
acquiring the last multi-hop information propagation vector of each neighborhood node associated with the target teacher, calculating the average value of the last multi-hop information propagation vector of each neighborhood node to obtain the last multi-hop information propagation vector of the neighborhood node, acquiring the last multi-hop information propagation vector of the target teacher, inputting the last multi-hop information propagation vector of the target teacher and the last multi-hop information propagation vector of the neighborhood node into a neural network, and calculating by the neural network to obtain the multi-hop information propagation vector of the target teacher.
The neighborhood nodes specifically refer to course nodes associated with teacher nodes, other nodes can be obtained through calculation through the method, and the specific designations of the neighborhood nodes are determined according to the multi-entity relation model.
The calculation formula of the multi-hop information propagation vector of the target student node is as follows:
wherein,propagation vector of kth hop information of target student node, W k And B k Are trainable parameter matrices, < >>Propagation vector for kth-1 hop of course-associated node,>a propagation vector of the kth-1 hop of the target student node a;
wherein,the kth-1 hop propagation vector of the course-associated node, N (a) is the course-associated node,>the propagation vector of the kth-1 hop for the jth course-associated node.
The method also comprises the step of updating the system parameters, and the method for updating the system parameters comprises the following steps:
acquiring a feedback value of a student actually on a recommended course in the last accumulated error back propagation and a feedback value of an expected student on the recommended course;
calculating accumulated errors of the obtained actual feedback value and the expected feedback value;
counter-propagating the accumulated errors, and completing updating of system parameters by a random gradient descent method;
the system parameters comprise a trainable parameter matrix, initial vectors of courses, initial vectors of teachers and initial vectors of schools.
The backward error propagation is completed through a mean square loss function, and the calculation formula of the backward error propagation is as follows:
wherein H is the system accumulated error, θ is the rewarding mapping vector, z is the hybrid characterization vector, L is the system recommended number between two accumulated error counter-propagation, r l Feedback for the first round of students on recommended courses,for the mixed characterization vector of students and courses in the first round, theta l-1 A bonus mapping vector for round l-1;
the update formula for the trainable parameter matrix is:
wherein W is k As a trainable parameter matrix, beta 1 For learning rate, H is system accumulated errorDifference;
wherein B is k As a trainable parameter matrix, beta 2 For learning rate, H is the accumulated error of the system;
the update formula for the course initial vector is:
wherein v is c Beta, the initial vector of course 3 For learning rate, H is the accumulated error of the system;
the update formula for the teacher initial vector is:
wherein v is t As initial vector of teacher, beta 4 For learning rate, H is the accumulated error of the system;
the update formula for the initial vector of the school is as follows:
wherein v is u As initial vector of school, beta 4 For learning rate, H is the system accumulated error.
As shown in FIG. 2, the system for recommending courses for students mainly comprises a multi-entity relationship construction module, an entity vector representation module and a course online recommendation module.
When the system operates, the entity vector representation module calculates and acquires the characterization vector of each node in the multi-entity relationship model according to the multi-entity relationship model constructed by the multi-entity relationship construction module, and the course online recommendation system recommends courses for students according to the characterization vector calculated by the entity vector representation module, and simultaneously acquires feedback of the students on the recommended courses during recommendation and updates the course recommendation model based on feedback information.
After the course recommendation model carries out recommendation for a certain number of times, error back propagation is carried out to update the initial vector of each node in the entity vector representation module, so that the recommendation of the system is gradually improved in the recommendation process.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (7)
1. An online course recommendation method, wherein the recommendation method comprises:
taking courses, students, schools and teachers as nodes in advance, and establishing an entity relation model of the courses about the students, schools and teachers respectively;
acquiring course association node information associated with a target student of a course to be recommended according to the established entity relation model;
calculating multi-hop information propagation vectors of the target students and each course association node, and taking the multi-hop information propagation vectors as characterization vectors of the target students and the course association nodes;
according to the characterization vector of the target student and the characterization vector of each course association node, the recommended course of the target student is completed;
the 0 th hop propagation vector of each node in the entity relation model is corresponding to the initial vector of each node in the entity relation model;
the characteristic vectors of the three entities of the course, the teacher and the school are respectively corresponding to an initial vector of the course, an initial vector of the teacher and an initial vector of the school;
the initial vector of the student is an average value of initial vectors of all courses selected by the student;
the calculation method of the multi-hop information propagation vector of the target student comprises the following steps:
acquiring last multi-hop information propagation vectors of all course association nodes associated with the target student node;
calculating the average value of the k-1 jump information propagation vectors of all course associated nodes to obtain the k-1 jump information propagation vector of the course associated node;
the method comprises the steps of obtaining a kth-1-hop information propagation vector of a target student node, and inputting the kth-1-hop information propagation vector of the target student node and a last-time multi-hop information propagation vector of a course association node into a neural network to obtain the kth-hop information propagation vector of the target student node;
the calculation formula of the multi-hop information propagation vector of the target student node is as follows:
wherein,propagation vector of kth hop information of target student node, W k And B k Are trainable parameter matrices, < >>Propagation vector for kth-1 hop of course-associated node,>a propagation vector of the kth-1 hop of the target student node a;
wherein,propagation vector of kth-1 hop of course association node, N (a) course association nodePoint (S)>The propagation vector of the kth-1 hop for the jth course-associated node.
2. The method of claim 1, further comprising obtaining feedback information of the student for recommending courses, calculating a bonus map vector of the course for the student based on the feedback information, and recommending the course for the student further comprises providing the bonus map vector.
3. The online course recommendation method of claim 2, wherein the bonus mapping vector update method is:
wherein: a is that l For updating the intermediate amount of the bonus map vector in the first round, T is the vector transpose of Z,a mixed characterization vector for the student and the recommended course in the first round;
wherein b l To update the intermediate quantity of the bonus map vector, r l Feedback for the first round of students on recommended courses,representing vectors for a mixture of students and courses in the first round;
wherein:representing vectors for a mix of students and courses in round I,>for the characterization vector of the first round student, +.>Representing a vector for a first course, wherein T is the transposition of the vector;
wherein: θ l For a round of bonus mapping vectors, A l Updating the intermediate quantity of the bonus map vector, b, for the first round l The intermediate amount of the bonus map vector is updated for the first round.
4. An online course recommendation method as claimed in claim 3, in which when said student selects recommended courses, then r l =1, when the student does not select recommended course, r l ＝0。
5. The online course recommendation method of claim 1, wherein the calculation formula for recommending courses for the student is:
wherein c l For the first round of courses recommended to students, C is a set of courses, θ is a bonus mapping vector, T is a vector transpose of Z,for a mixed token vector of students and courses, α is the exploration factor, and a is the intermediate quantity of updating the bonus map vector.
6. The online course recommendation method of claim 1, wherein a feedback value of the actual student to the recommended course and a feedback value of the expected student to the recommended course in the last cumulative error back propagation are obtained;
calculating accumulated errors of the obtained actual feedback value and the expected feedback value;
counter-propagating the accumulated errors, and completing updating of system parameters by a random gradient descent method;
wherein the system parameters include a trainable parameter matrix, an initial vector of a course, an initial vector of a teacher, and an initial vector of a school.
7. The online class-recommendation method of claim 6, wherein the reverse error propagation is performed by a mean square loss function, and wherein the reverse error propagation is calculated by the formula:
wherein H is the system accumulated error, θ is the rewarding mapping vector, z is the hybrid characterization vector, L is the system recommended number between two accumulated error counter-propagation, r l Feedback for the first round of students on recommended courses,for the mixed characterization vector of students and courses in the first round, theta l-1 A bonus mapping vector for round l-1;
the update formula for the trainable parameter matrix is:
wherein W is k As a trainable parameter matrix, beta 1 For learning rate, H is the accumulated error of the system;
wherein B is k As a trainable parameter matrix, beta 2 For learning rate, H is the accumulated error of the system;
the update formula for the course initial vector is:
wherein v is c Beta, the initial vector of course 3 For learning rate, H is the accumulated error of the system;
the updating formula of the initial vector of the teacher is as follows:
wherein v is t As initial vector of teacher, beta 4 For learning rate, H is the accumulated error of the system;
the update formula for the initial vector of the school is as follows:
wherein v is u As initial vector of school, beta 4 For learning rate, H is the system accumulated error.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111011484.6A CN113722591B (en) | 2021-08-31 | 2021-08-31 | Online course recommendation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111011484.6A CN113722591B (en) | 2021-08-31 | 2021-08-31 | Online course recommendation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113722591A CN113722591A (en) | 2021-11-30 |
| CN113722591B true CN113722591B (en) | 2024-02-23 |
Family
ID=78679616
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111011484.6A Active CN113722591B (en) | 2021-08-31 | 2021-08-31 | Online course recommendation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113722591B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460249A (en) * | 2020-02-24 | 2020-07-28 | 桂林电子科技大学 | Personalized learning resource recommendation method based on learner preference modeling |
| WO2020192567A1 (en) * | 2019-03-22 | 2020-10-01 | 深圳市元征科技股份有限公司 | Teaching management method and system, block chain node device and computer storage medium |
| US10878710B1 (en) * | 2014-09-03 | 2020-12-29 | Eab Global, Inc. | Systems and methods for electronic platform and data processing for student success in higher education institutions |
| CN112732892A (en) * | 2020-12-30 | 2021-04-30 | 平安科技（深圳）有限公司 | Course recommendation method, device, equipment and storage medium |
-
2021
- 2021-08-31 CN CN202111011484.6A patent/CN113722591B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10878710B1 (en) * | 2014-09-03 | 2020-12-29 | Eab Global, Inc. | Systems and methods for electronic platform and data processing for student success in higher education institutions |
| WO2020192567A1 (en) * | 2019-03-22 | 2020-10-01 | 深圳市元征科技股份有限公司 | Teaching management method and system, block chain node device and computer storage medium |
| CN111460249A (en) * | 2020-02-24 | 2020-07-28 | 桂林电子科技大学 | Personalized learning resource recommendation method based on learner preference modeling |
| CN112732892A (en) * | 2020-12-30 | 2021-04-30 | 平安科技（深圳）有限公司 | Course recommendation method, device, equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于注意力的深度协同在线学习资源推荐模型;冯金慧;陶宏才;;成都信息工程大学学报(02);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113722591A (en) | 2021-11-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108614865B (en) | Personalized learning recommendation method based on deep reinforcement learning | |
| CN110458663A (en) | A kind of vehicle recommended method, device, equipment and storage medium | |
| CN108304489A (en) | A kind of goal directed type personalization dialogue method and system based on intensified learning network | |
| Kremer-Hayon et al. | Themes, processes and trends in the professional development of teacher educators | |
| CN109191345A (en) | A kind of cognitive diagnosis method of Student oriented cognitive process | |
| CN113722591B (en) | Online course recommendation method | |
| CN114937366B (en) | Traffic flow calculation method based on multi-scale traffic demand and supply conversion | |
| CN111752304A (en) | Unmanned aerial vehicle data acquisition method and related equipment | |
| CN108229688A (en) | A kind of information processing method and device based on IRT | |
| CN115545587A (en) | Classroom teaching scheme evaluation method and system based on multiple agents | |
| CN116319355A (en) | Satellite network topology generation method based on deep reinforcement learning | |
| CN113870634A (en) | Intelligent volume combination method and system combined with virtual teaching | |
| CN207020711U (en) | A kind of Classroom Teaching device | |
| CN109345048A (en) | Prediction technique, device, electronic equipment and computer readable storage medium | |
| CN114943278B (en) | Continuous online group incentive method and device based on reinforcement learning and storage medium | |
| CN110302539A (en) | A kind of tactics of the game calculation method, device, system and readable storage medium storing program for executing | |
| CN109918406A (en) | A kind of matter group volume method and papers generation system such as automatic based on maximum clique algorithm | |
| CN108009150A (en) | A kind of input method and device based on Recognition with Recurrent Neural Network | |
| CN110727822B (en) | Online learning system based on personalized recommendation | |
| CN110675279A (en) | Power grid building simulation system based on AI visualization | |
| CN109166372A (en) | Based on the tutoring system and decision-making technique under cloud framework | |
| Divina et al. | A hybrid model for learning word-meaning mappings | |
| CN110767018B (en) | Microwave teaching system for automatically configuring parameters and working method thereof | |
| CN114844789B (en) | Community knowledge sharing evaluation method based on evolution game model | |
| Alan et al. | Using bayesian networks for knowledge representation and evaluation in intelligent tutoring systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |