CN111414540A

CN111414540A - Online learning recommendation method and device, online learning system and server

Info

Publication number: CN111414540A
Application number: CN202010198632.9A
Authority: CN
Inventors: 张明
Original assignee: Individual
Current assignee: Chongqing Tancheng Digital Technology Group Co ltd
Priority date: 2020-03-20
Filing date: 2020-03-20
Publication date: 2020-07-14
Anticipated expiration: 2040-03-20
Also published as: CN112287227A; CN112287228A; CN111414540B

Abstract

The embodiment of the application provides an online learning recommendation method, an online learning recommendation device, an online learning system and a server, when learning behavior data of a student user and all online learning recommendation models do not exist and reach a preset adaptation threshold, a plurality of model matching attributes corresponding to the student user and model attribute nodes loaded in each model matching attribute are matched by combining recommendation levels of all online learning recommendation models relative to the student user, so that style attribute weight parameters corresponding to each model matching attribute are associated to second target online learning recommendation models corresponding to the model attribute nodes loaded in each model matching attribute, information recommendation is carried out on the student user through each second target online learning recommendation model according to the style attribute weight parameters, recommendation precision can be improved in real time, and information recommendation caused by the fact that the parameter configuration process of the online learning recommendation models cannot be effectively updated and continuously carried out in a short time is avoided Experience is poor.

Description

Online learning recommendation method and device, online learning system and server

Technical Field

The application relates to the technical field of big data analysis and information recommendation, in particular to an online learning recommendation method and device, an online learning system and a server.

Background

With the rapid development of internet technology, a plurality of student users can learn widely through the internet by establishing an online learning platform on the internet. In a conventional scheme, in order to improve the learning experience of a student user, an online learning platform usually collects learning behaviors of the student user in a learning process, and performs big data analysis on the learning behaviors and then matches the learning behaviors with a plurality of online learning recommendation models to perform subsequent online learning recommendation on online learning recommendation models corresponding to the association of the student user.

The research of the inventor of the application finds that for an online learning platform, the built online learning recommendation model is usually limited, and the online learning recommendation model with the best matching degree is preferably associated to the corresponding student user in the actual model matching process. However, when the matching degree is low (lower than a certain matching degree threshold), even if the online learning recommendation model with the best matching degree is adopted, it is difficult to achieve a more accurate recommendation precision, and the parameter configuration process of the online learning recommendation model cannot be effectively updated in a short time, so that the information recommendation experience is continuously poor.

Disclosure of Invention

In view of this, an object of the present application is to provide an online learning recommendation method, an online learning recommendation device, an online learning system, and a server, which can improve recommendation accuracy in real time, and avoid a situation of poor information recommendation experience caused by a parameter configuration process of an online learning recommendation model being unable to be updated effectively and continuously in a short time.

In a first aspect, the present application provides an online learning recommendation method, which is applied to a server, where the server is in communication connection with a plurality of online learning terminals, and the method includes:

acquiring learning behavior data corresponding to a student user from each online learning terminal, and extracting learning behavior style data corresponding to the learning behavior data, wherein the learning behavior data are acquired by the server according to online learning mode information of the online learning terminal corresponding to the student user and an online learning interaction mode between the online learning terminal and the server;

acquiring a learning style weight parameter of an online learning recommendation model of each student user and a recommendation weight corresponding to a recommendation strategy of each online learning recommendation model under the corresponding learning style weight parameter, and calculating the adaptation degree between the learning behavior data and each online learning recommendation model according to a style weight corresponding to style data in the learning behavior style data, a learning style weight parameter of each online learning recommendation model and a recommendation weight corresponding to the learning style weight parameter in the recommendation strategy under the corresponding learning style weight parameter;

when the target adaptation degree reaching a preset adaptation degree threshold value exists in all the determined adaptation degrees, associating the student user to a first target online learning recommendation model corresponding to the target adaptation degree so that the first target online learning recommendation model carries out information recommendation on the student user to obtain a first online learning recommendation result;

when the target adaptation degree which reaches the preset adaptation degree threshold value does not exist in all the determined adaptation degrees, determining the recommendation level of each online learning recommendation model relative to the student user, performing data feature recognition on the learning behavior data of the student user through each corresponding online learning recommendation model according to the magnitude sequence of the recommendation levels to obtain a plurality of model matching attributes corresponding to the student user and a model attribute node corresponding to the recommendation level of each online learning recommendation model in each model matching attribute, determining a style attribute weight parameter corresponding to each model matching attribute from the learning behavior style data corresponding to the student user according to each model matching attribute, and associating the style attribute weight parameter corresponding to each model matching attribute to a second target online learning recommendation model corresponding to the model attribute node loaded in each model matching attribute, and performing information recommendation on the student user through each second target online learning recommendation model according to the style attribute weight parameters to obtain a second online learning recommendation result.

In a possible design of the first aspect, the step of calculating the degree of adaptation between the learning behavior data and each online learning recommendation model according to the style weight corresponding to the style data in the learning behavior style data, the learning style weight parameter of each online learning recommendation model, and the recommendation weight corresponding to the learning style weight parameter in the recommendation policy under the corresponding learning style weight parameter includes:

calculating style conversion weights of style weights corresponding to style data in the learning behavior style data under the learning style weight parameters of each online learning recommendation model;

calculating the absolute value of the weight difference between each style conversion weight and the corresponding recommendation weight in the recommendation strategy under the corresponding learning style weight parameter;

and adding the absolute values of the calculated weight difference values to obtain the adaptation degree between the learning behavior data and each online learning recommendation model.

In a possible design of the first aspect, the step of determining a recommendation level of each online learning recommendation model with respect to the trainee user includes:

the method comprises the steps of obtaining a plurality of model weight nodes corresponding to each online learning recommendation model, determining a content feature sequence corresponding to recommended content features in each model weight node and a plurality of content influence parameters, wherein the content feature sequence is used for representing recommendation tendency behaviors of the recommended content features in a learning recommendation process, and the content influence parameters are used for representing influence weights of the recommended content features on the learning recommendation process;

when a first style dimension characteristic sequence is determined to be contained in each model weight node according to the content characteristic sequence, according to content influence parameters and position information of each model weight node under the first style dimension characteristic sequence, determining first matching parameters between each content influence parameter of each model weight node under a second style dimension characteristic sequence and each content influence parameter of each model weight node under the first style dimension characteristic sequence, wherein the first style dimension characteristic sequence represents a characteristic sequence of a learning content style, and the second style dimension characteristic sequence represents a characteristic sequence of a learning behavior style;

when the fact that each model weight node comprises a first style dimension characteristic sequence is determined according to the content characteristic sequence, according to content influence parameters and position information of each model weight node under the first style dimension characteristic sequence, first matching parameters between each content influence parameter of each model weight node under a second style dimension characteristic sequence and each content influence parameter of each model weight node under the first style dimension characteristic sequence are determined;

transferring content influence parameters of which the first matching parameters between the content influence parameters of the model weight nodes under the second style dimension characteristic sequence and the first style dimension characteristic sequence reach a preset parameter range to the first style dimension characteristic sequence;

when each model weight node comprises a plurality of content influence parameters under the second style dimension characteristic sequence, determining second matching parameters among the content influence parameters of each model weight node under the second style dimension characteristic sequence according to the content influence parameters and the position information of each model weight node under the first style dimension characteristic sequence, and screening the content influence parameters under the second style dimension characteristic sequence according to the second matching parameters among the content influence parameters;

setting a list position grade for the screened target content influence parameters according to the content influence parameters and the position information of each model weight node under the first style dimension characteristic sequence, and transferring the target content influence parameters to a list interval corresponding to the list position grade in the first style dimension characteristic sequence;

and after weighting processing is respectively carried out according to all the content influence parameters in the first style dimension characteristic sequence, multiplying the weighted parameters by the adaptation degrees corresponding to the corresponding online learning recommendation models, and determining the recommendation level of the online learning recommendation models corresponding to the model weight nodes relative to the student user.

In a possible design of the first aspect, the step of performing data feature recognition on the trainee user according to the size order of the recommendation level to obtain a plurality of model matching attributes corresponding to the trainee user and a model attribute node loaded in each model matching attribute and corresponding to the recommendation level of each online learning recommendation model includes:

listing recommended content nodes corresponding to each recommendation level, and establishing an online learning recommendation model node sequence, wherein the online learning recommendation model node sequence is a itemized processing list, each model region corresponds to one group of sequence features, each group of sequence features is provided with at least one recommended content node, and each model region of the online learning recommendation model node sequence has a progressive relation from high to low;

determining preset configuration attribute information of the student user, and extracting recommended content nodes in at least one online learning recommendation model node sequence contained in the preset configuration attribute information of the student user;

establishing a mapping relation between the recommended content node and the online learning recommendation model node sequence, and generating a mapping recommendation strategy according to the mapping relation;

generating a mapping recommendation strategy according to the mapping relation, wherein the mapping recommendation strategy comprises the following steps: converting the corresponding online learning recommendation model node sequence into model node data according to each recommendation content node, respectively generating at least one recommendation unit of each model node data, then obtaining the recommendation units with different recommendation levels to form a recommendation unit group, mapping each recommendation unit in the recommendation unit group into the online learning recommendation model node sequence to form a mapping recommendation strategy, wherein each recommendation unit is in one-to-one correspondence with one recommendation item content;

traversing and comparing recommended content nodes contained in the preset configuration attribute information of the student user with each recommended content node in the mapping recommendation strategy, and recording a recommendation unit as a model matching attribute direction of the student user if all the recommended content nodes of the recommendation unit are contained in the preset configuration attribute information of the student user in the traversing and comparing process;

and determining a plurality of matching positions corresponding to the student user according to the model matching attribute directions of the student user, performing data feature identification on the student user according to each matching position to obtain a corresponding model matching attribute, and determining a model attribute node of a recommended level according to at least one recommended unit of loaded model node data of the recommended level included in each model matching attribute.

In a possible design of the first aspect, the step of determining preset configuration attribute information of the trainee user includes:

performing itemized processing on the student user to obtain a plurality of pieces of registration project information based on a plurality of pieces of user registration data information formed by the registration node of the student user and the currently used node of the student user, which are stored in the server and correspond to the student user;

acquiring information before project editing and information after project editing of each registered project information;

establishing a project editing behavior set of project editing records corresponding to the student user according to the information before project editing and the information after project editing of each piece of registered project information;

acquiring a plurality of editing attribute units corresponding to the project editing record corresponding to the student user, and counting target editing attribute units in the editing attribute units, wherein the target editing attribute units have editing character code characteristics;

judging whether an associated editing item exists between two adjacent target editing attribute units, if so, counting the number of the associated editing items, implanting the item editing behavior set into each user registration data information when the number does not exceed a set value, acquiring an updated item editing behavior set when the item editing behavior set implanted into each user registration data information is updated, and counting editing behavior characteristics and registered item information deviation information corresponding to each acquired updated item editing behavior set;

determining the editing weight of each updated project editing behavior set according to the editing behavior characteristics corresponding to each updated project editing behavior set and the registered project information deviation information;

and correcting the project editing behavior set which is obtained in real time and is updated according to the editing weight to obtain a student user block editing sequence, extracting learning behavior style data in the registered data information of each user according to sequence characteristics in the student user block editing sequence, and determining preset configuration attribute information of the student user according to the extracted learning behavior style data.

In a possible design of the first aspect, the step of performing data feature recognition on the trainee user according to each matching position to obtain a corresponding model matching attribute includes:

acquiring current behavior characteristics of the student user and positioning first model characteristics corresponding to each matching position from the current behavior characteristics;

judging whether a first model feature corresponding to each matching position in the current behavior features has a matched feature value relative to a second model feature in the current behavior features, wherein the second model feature is a feature except the first model feature in the current behavior features;

if so, determining the first model characteristic corresponding to each matching position located from the current behavior characteristics as the effective model characteristic of the current behavior characteristics, otherwise, performing weighted summation on the first model characteristic corresponding to each matching position located from the current behavior characteristics and the second model characteristic in the current behavior characteristics, and determining the weighted summation result as the effective model characteristic of the current behavior characteristics;

aiming at each matching position, extracting a first model matching command line implanted into the running thread of the server from the matching position, and fusing partial characteristics in the effective model characteristics of the current behavior characteristics with the first model matching command line to obtain a second model matching command line;

respectively operating the first model matching command line and the second model matching command line in the mirror image thread corresponding to the operating thread to obtain a first operating result and a second operating result which respectively correspond to the first operating result and the second operating result;

judging whether the similarity of the first operation result and the second operation result reaches a preset threshold value, starting the matching position to operate the second model matching command line when the similarity of the first operation result and the second operation result reaches the preset threshold value, obtaining a third operation result corresponding to the second model matching command line, extracting feature classification information in the third operation result, obtaining a model matching attribute corresponding to the matching position according to the feature classification information, and returning to the step of fusing partial features in the effective model features of the current behavior features with the first model matching command line to obtain the second model matching command line when the similarity of the first operation result and the second operation result does not reach the preset threshold value.

In one possible design of the first aspect, the method further includes:

obtaining a second online learning recommendation result obtained by each second target online learning recommendation model recommending information to the student user according to the style attribute weight parameters;

obtaining a recommendation style characteristic of a recommendation style corresponding to each second online learning recommendation result according to the recommendation content item of each second online learning recommendation result;

obtaining a recommendation screening result according to the feature screening range of the preset recommendation style of the student user and the recommendation style feature of the recommendation style corresponding to the second online learning recommendation result, wherein the recommendation screening result comprises a plurality of style feature sets corresponding to the recommendation style features in the feature screening range of the preset recommendation style;

obtaining screening feature content information of any one first screening feature content with different style features contained in the recommended screening result, determining screening feature content attributes of the first screening feature content according to the screening feature content information of the first screening feature content, and determining a target learning scene corresponding to the first screening feature content based on a screening feature content scene in the screening feature content information of the first screening feature content;

determining recommended course information matched with the screening characteristic content attribute of the first screening characteristic content, and selecting recommended courses matched with the recommended course information;

according to the screening characteristic content attribute of the first screening characteristic content and the class labels of a plurality of recommended classes with the recommended class information in the target learning scene, selecting a target recommended class matched with the first screening characteristic content from the plurality of recommended classes with the recommended class information, wherein the target recommended class is also required to be matched with a second screening characteristic content associated with the first screening characteristic content;

obtaining course label information of the first screening characteristic content included in the screening characteristic content information of the first screening characteristic content, and obtaining course label information of the second screening characteristic content included in the screening characteristic content information of the second screening characteristic content;

and generating a corresponding third online learning recommendation result according to the course label information of the first screening characteristic content and the course label information of the second screening characteristic content.

In a second aspect, an embodiment of the present application further provides an online learning recommendation device, which is applied to a server, where the server is in communication connection with a plurality of online learning terminals, and the device includes:

the extraction module is used for acquiring learning behavior data corresponding to a student user from each online learning terminal and extracting learning behavior style data corresponding to the learning behavior data, wherein the learning behavior data are obtained by the server according to online learning mode information of the online learning terminal corresponding to the student user and an online learning interaction mode between the online learning terminal and the server;

the calculation module is used for acquiring a learning style weight parameter of an online learning recommendation model of each student user and a recommendation weight corresponding to a recommendation strategy of each online learning recommendation model under the corresponding learning style weight parameter, and calculating the adaptation degree between the learning behavior data and each online learning recommendation model according to the style weight corresponding to the style data in the learning behavior style data, the learning style weight parameter of each online learning recommendation model and the recommendation weight corresponding to the learning style weight parameter in the recommendation strategy under the corresponding learning style weight parameter;

the first recommendation module is used for associating the student user to a first target online learning recommendation model corresponding to a preset adaptation degree when the target adaptation degree reaching a preset adaptation degree threshold exists in all the determined adaptation degrees so that the first target online learning recommendation model carries out information recommendation on the student user to obtain a first online learning recommendation result;

a second recommending module, configured to, when there is no target adaptation degree that reaches the preset adaptation degree threshold in all the determined adaptation degrees, determine a recommendation level of each online learning recommending model, perform data feature recognition on the learning behavior data of the learner user through each corresponding online learning recommending model according to a magnitude order of the recommendation levels to obtain a plurality of model matching attributes corresponding to the learner user and a model attribute node corresponding to the recommendation level of each online learning recommending model in each model matching attribute, determine a style attribute weight parameter corresponding to each model matching attribute from the learning behavior style data corresponding to the learner user according to each model matching attribute, and associate the style attribute weight parameter corresponding to each model matching attribute to a second target online learning recommending model corresponding to the model attribute node loaded in each model matching attribute, and performing information recommendation on the student user through each second target online learning recommendation model according to the style attribute weight parameters to obtain a second online learning recommendation result.

In a third aspect, an embodiment of the present application further provides an online learning system, where the online learning system includes a server and a plurality of online learning terminals communicatively connected to the server;

each online learning terminal is used for sending learning behavior data corresponding to the college users to the server;

the server is used for acquiring learning behavior data corresponding to a student user from each online learning terminal and extracting learning behavior style data corresponding to the learning behavior data, wherein the learning behavior data are obtained by the server according to online learning mode information of the online learning terminal corresponding to the student user and an online learning interaction mode between the online learning terminal and the server;

the server is used for acquiring learning style weight parameters of online learning recommendation models of each student user and recommendation weights corresponding to recommendation strategies of each online learning recommendation model under the corresponding learning style weight parameters, and calculating the adaptation degree between the learning behavior data and each online learning recommendation model according to style weights corresponding to style data in the learning behavior style data, learning style weight parameters of each online learning recommendation model and recommendation weights corresponding to the learning style weight parameters in the recommendation strategies under the corresponding learning style weight parameters;

when the target adaptation degree reaching a preset adaptation degree threshold value exists in all the determined adaptation degrees, the server is used for associating the student user to a first target online learning recommendation model corresponding to the target adaptation degree so that the first target online learning recommendation model carries out information recommendation on the student user to obtain a first online learning recommendation result;

when the target adaptation degree which reaches the preset adaptation degree threshold value does not exist in all the determined adaptation degrees, the server is used for determining the recommendation level of each online learning recommendation model, performing data feature recognition on the learning behavior data of the student user through each corresponding online learning recommendation model according to the magnitude sequence of the recommendation level to obtain a plurality of model matching attributes corresponding to the student user and a model attribute node corresponding to the recommendation level of each online learning recommendation model in each model matching attribute, determining a style attribute weight parameter corresponding to each model matching attribute from the learning behavior style data corresponding to the student user according to each model matching attribute, and associating the style attribute weight parameter corresponding to each model matching attribute to a second target online learning recommendation model corresponding to the model attribute node loaded in each model matching attribute, and performing information recommendation on the student user through each second target online learning recommendation model according to the style attribute weight parameters to obtain a second online learning recommendation result.

In a fourth aspect, an embodiment of the present application further provides a server, where the server includes a processor, a machine-readable storage medium, and a network interface, where the machine-readable storage medium, the network interface, and the processor are connected through a bus system, the network interface is configured to be communicatively connected to at least one online learning terminal, the machine-readable storage medium is configured to store a program, an instruction, or a code, and the processor is configured to execute the program, the instruction, or the code in the machine-readable storage medium to perform the online learning recommendation method in the first aspect or any one of the possible designs in the first aspect.

In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, where instructions are stored, and when executed, cause a computer to perform the online learning recommendation method in the first aspect or any one of the possible designs of the first aspect.

According to any one of the aspects, by combining learning behavior style data corresponding to learning behavior data of a student user, when the learning behavior style data does not exist between the online learning recommendation models and reach a preset adaptation threshold, combining a plurality of model matching attributes corresponding to the student user and model attribute nodes loaded in each model matching attribute of each online learning recommendation model relative to the recommendation level of the student user, so that style attribute weight parameters corresponding to each model matching attribute are associated to second target online learning recommendation models corresponding to the model attribute nodes loaded in each model matching attribute, and information recommendation is performed on the student user through each second target online learning recommendation model according to the style attribute weight parameters, so that the recommendation precision can be improved in real time, and the problem that the information recommendation experience is relatively high due to the fact that the parameter configuration process of the online learning recommendation models cannot be effectively updated and continuously performed in a short time is avoided And a poor case.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 is a schematic view of an application scenario of an online learning system according to an embodiment of the present application;

fig. 2 is a schematic flowchart of an online learning recommendation method according to an embodiment of the present application;

fig. 3 is a schematic functional module diagram of an online learning recommendation device according to an embodiment of the present application;

fig. 4 is a block diagram schematically illustrating a structure of a server for implementing the online learning recommendation method according to an embodiment of the present application.

Detailed Description

The present application will now be described in detail with reference to the drawings, and the specific operations in the method embodiments may also be applied to the apparatus embodiments or the system embodiments.

Fig. 1 is an interaction diagram of an online learning system 10 according to an embodiment of the present application. The online learning system 10 may include a server 100 and an online learning terminal 200 communicatively connected to the server 100 via a network, the online learning system 10 shown in fig. 1 is only one possible example, and in other possible embodiments, the online learning system 10 may include only a part of the components shown in fig. 1 or may also include other components.

In this embodiment, the online learning terminal 200 may include a mobile device, a tablet computer, a laptop computer, etc., or any combination thereof. In some embodiments, the mobile device may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the smart home devices may include control devices of smart electrical devices, smart monitoring devices, smart televisions, smart cameras, and the like, or any combination thereof. In some embodiments, the wearable device may include a smart bracelet, a smart lace, smart glass, a smart helmet, a smart watch, a smart garment, a smart backpack, a smart accessory, or the like, or any combination thereof. In some embodiments, the smart mobile device may include a smartphone, a personal digital assistant, a gaming device, and the like, or any combination thereof. In some embodiments, the virtual reality device and/or the augmented reality device may include a virtual reality helmet, virtual reality glass, a virtual reality patch, an augmented reality helmet, augmented reality glass, an augmented reality patch, or the like, or any combination thereof. For example, the virtual reality device and/or augmented reality device may include various virtual reality products and the like.

In order to solve the technical problem in the foregoing background art, fig. 2 is a flowchart illustrating an online learning recommendation method according to an embodiment of the present application, which may be executed by the server 100 shown in fig. 1, and the online learning recommendation method is described in detail below.

Step S110, obtaining learning behavior data corresponding to the learner user from each online learning terminal 200, and extracting learning behavior style data corresponding to the learning behavior data.

In this embodiment, the learning behavior data may be obtained by the server 100 according to online learning mode information of the online learning terminal 200 corresponding to the learner user and an online learning interaction mode between the online learning terminal 200 and the server 100. For example, the online learning mode information may be used to represent online learning modes of the student user, such as a pupil learning mode, a middle school student learning mode, and a college student learning mode, or a pre-study learning mode, a review learning mode, and a study mode for making an examination, and for different online learning modes of the user, the learning behavior data may be acquired according to different strategies, and may be flexibly adjusted according to actual design requirements, which is not limited herein. For another example, the online learning interactive mode may refer to an interactive mode between the learner user and the online learning platform (server 100), such as a one-way question-answer interactive mode, a two-way question-answer interactive mode, and the like, which is not limited herein.

Step S120, obtaining the learning style weight parameter of the online learning recommendation model of each student user and the recommendation weight corresponding to the recommendation strategy of each online learning recommendation model under the corresponding learning style weight parameter, and calculating the adaptation degree between the learning behavior data and each online learning recommendation model according to the style weight corresponding to the style data in the learning behavior style data, the learning style weight parameter of each online learning recommendation model and the recommendation weight corresponding to the recommendation strategy under the corresponding learning style weight parameter.

In this embodiment, the learning style weight parameter may refer to a weight parameter type occupied by a specific learning style, such as a music style weight parameter, a mathematic style weight parameter, and the like, and recommendation strategies corresponding to different learning style weight parameters are different, and for different online learning recommendation models, recommendation weights corresponding to recommendation strategies corresponding to different learning style weight parameters are also different, and may be specifically configured and trained in advance, which is not described in detail herein.

Step S130, when the target adaptation degree reaching the preset adaptation degree threshold exists in all the determined adaptation degrees, associating the student user to a first target online learning recommendation model corresponding to the target adaptation degree so that the first target online learning recommendation model carries out information recommendation on the student user to obtain a first online learning recommendation result.

Step S140, when there is no target adaptation degree reaching the preset adaptation degree threshold in all the determined adaptation degrees, determining the recommendation level of each online learning recommendation model relative to the student user and according to the magnitude sequence of the recommendation levels, performing data feature recognition on the learning behavior data of the student user through each corresponding online learning recommendation model to obtain a plurality of model matching attributes corresponding to the student user and a model attribute node corresponding to the recommendation level of each online learning recommendation model in each model matching attribute, determining a style attribute weight parameter corresponding to each model matching attribute from the learning behavior style data corresponding to the student user according to each model matching attribute, and associating the style attribute weight parameter corresponding to each model matching attribute to a second target online learning recommendation model corresponding to the model attribute node loaded in each model matching attribute, and performing information recommendation on the student users through each second target online learning recommendation model according to the style attribute weight parameters to obtain a second online learning recommendation result.

Based on the above design, the present embodiment can determine, when the non-existence between the learning behavior data of the learner user and each online learning recommendation model reaches the preset fitness threshold, in combination with the plurality of model matching attributes corresponding to the student user and the model attribute nodes loaded in each model matching attribute, which are matched by each online learning recommendation model relative to the recommendation level of the student user, thereby associating the style attribute weight parameter corresponding to each model matching attribute to the second target online learning recommendation model corresponding to the model attribute node loaded in each model matching attribute, information recommendation is carried out on the student users according to the style attribute weight parameters through each second target online learning recommendation model, therefore, the recommendation precision can be improved in real time, and the condition of poor information recommendation experience caused by the fact that the parameter configuration process of the online learning recommendation model cannot be effectively updated continuously in a short time is avoided.

In one possible design, for step S120, the present embodiment may calculate a style conversion weight of a style weight corresponding to style data in the learning behavior style data under a learning style weight parameter of each online learning recommendation model, and then calculate an absolute value of a weight difference between each style conversion weight and a corresponding recommendation weight in a recommendation policy under the corresponding learning style weight parameter, thereby adding the calculated absolute values of the weight differences to obtain a degree of adaptation between the learning behavior data and each online learning recommendation model.

For example, assuming that the learning style weight parameters of the online learning recommendation model a include a weight parameter a1, a weight parameter a2, and a weight parameter A3, and the style weight corresponding to the style data in the learning behavior style data is B, the style conversion weight B1, the style conversion weight B2, and the style conversion weight B3 of the style weight B under the weight parameter a1, the weight parameter a2, and the weight parameter A3, respectively, may be calculated, and then the fitness between the learning behavior data and the online learning recommendation model a may be calculated as | a1-a2| + | a1-a2| + | a1-a2 |.

In a possible design, based on the foregoing description, when there is a target adaptation degree reaching a preset adaptation degree threshold in all the determined adaptation degrees, the trainee user may be associated to a first target online learning recommendation model corresponding to the target adaptation degree, so that the first target online learning recommendation model performs information recommendation on the trainee user to obtain a first online learning recommendation result. The key point of the present application is how to improve the recommendation accuracy in real time when the target adaptation degree that reaches the preset adaptation degree threshold does not exist in all the determined adaptation degrees, and a situation that the information recommendation experience is poor due to the fact that the parameter configuration process of the online learning recommendation model cannot be updated effectively and continuously in a short time is avoided, so the following embodiment will focus on the detailed explanation of step S140.

For example, in a possible design, in step S140, in the process of determining the recommendation level of each online learning recommendation model with respect to the student user, the embodiment may obtain a plurality of model weight nodes corresponding to each online learning recommendation model, and determine a content feature sequence corresponding to the recommended content feature in each model weight node and a plurality of content influence parameters.

It should be noted that the content feature sequence may be used to characterize recommendation tendency behavior of recommended content features in the learning recommendation process, and the content influence parameter may be used to characterize influence weights of the recommended content features in the learning recommendation process.

On this basis, when it is determined that each model weight node includes the first style dimension feature sequence according to the content feature sequence, a first matching parameter between each content influence parameter of each model weight node under the second style dimension feature sequence and each content influence parameter of each model weight node under the first style dimension feature sequence is determined according to the content influence parameter and the position information of each model weight node under the first style dimension feature sequence.

It should be noted that the first style dimension feature sequence may represent a feature sequence of a learning content style, and the second style dimension feature sequence may represent a feature sequence of a learning behavior style.

Therefore, when the first style dimension characteristic sequence is determined to be contained in each model weight node according to the content characteristic sequence, a first matching parameter between each content influence parameter of each model weight node under the second style dimension characteristic sequence and each content influence parameter of each model weight node under the first style dimension characteristic sequence can be determined according to the content influence parameter and the position information of each model weight node under the first style dimension characteristic sequence, and then the content influence parameter of each model weight node under the second style dimension characteristic sequence and the content influence parameter under the first style dimension characteristic sequence, wherein the first matching parameter between the content influence parameters of each model weight node under the second style dimension characteristic sequence and under the first style dimension characteristic sequence reaches the preset parameter range, is transferred to the first style dimension characteristic sequence. And then, when each model weight node comprises a plurality of content influence parameters under the second style dimension characteristic sequence, determining second matching parameters among the content influence parameters of each model weight node under the second style dimension characteristic sequence according to the content influence parameters and the position information of each model weight node under the first style dimension characteristic sequence, and screening the content influence parameters under the second style dimension characteristic sequence according to the second matching parameters among the content influence parameters.

In this way, a list position grade can be set for the target content influence parameter obtained by screening according to the content influence parameter and the position information of each model weight node under the first style dimension characteristic sequence, and the target content influence parameter is transferred to a list interval corresponding to the list position grade in the first style dimension characteristic sequence, so that the recommendation grade of the online learning recommendation model corresponding to each model weight node relative to the student user is determined by multiplying the adaptation degree corresponding to the corresponding online learning recommendation model after weighting all the content influence parameters in the first style dimension characteristic sequence.

For another example, in a possible design, in step S140, in a process of performing data feature identification on the trainee user according to the size sequence of the recommendation levels to obtain a plurality of model matching attributes corresponding to the trainee user and a model attribute node loaded in each model matching attribute and corresponding to the recommendation level of each online learning recommendation model, this embodiment may list recommendation content nodes corresponding to each recommendation level, and establish an online learning recommendation model node sequence.

Optionally, the online learning recommendation model node sequence may be a itemized processing list, each model region corresponds to one group of sequence features, each group of sequence features has at least one recommendation content node, and each model region of the online learning recommendation model node sequence has a progressive relationship from high to low.

Meanwhile, in this embodiment, preset configuration attribute information of the trainee user may be determined, and recommended content nodes in at least one online learning recommendation model node sequence included in the preset configuration attribute information of the trainee user are extracted.

Then, a mapping relation between the recommended content node and the online learning recommendation model node sequence can be established, and a mapping recommendation strategy is generated according to the mapping relation.

For example, generating a mapping recommendation policy according to the mapping relationship may specifically be: converting the corresponding online learning recommendation model node sequence into model node data according to each recommendation content node, respectively generating at least one recommendation unit of each model node data, then obtaining recommendation units with different recommendation levels to form a recommendation unit group, mapping each recommendation unit in the recommendation unit group into the online learning recommendation model node sequence to form a mapping recommendation strategy, wherein each recommendation unit is in one-to-one correspondence with one recommendation item content.

Therefore, the recommended content nodes contained in the preset configuration attribute information of the student user can be subjected to traversal comparison with the recommended content nodes in the mapping recommendation strategy, in the traversal comparison process, if all the recommended content nodes of one recommendation unit are contained in the preset configuration attribute information of the student user, the recommendation unit is recorded as the model matching attribute direction of the student user, then a plurality of matching positions corresponding to the student user are determined according to the model matching attribute directions of the student user, the corresponding model matching attributes are obtained by performing data feature recognition on the student user according to each matching position, and the model attribute nodes of the recommendation level are determined according to at least one recommendation unit of the loaded model node data of the recommendation level contained in each model matching attribute.

In one possible example, in the process of determining the preset configuration attribute information of the trainee user, the present embodiment may perform item splitting processing on the trainee user to obtain a plurality of pieces of registered item information based on a plurality of pieces of user registered material information formed by the registered node of the trainee user stored in the server 100 corresponding to the trainee user and the currently used node of the trainee user, acquire pre-item editing information and post-item editing information of each piece of registered item information, and then establish a set of item editing behaviors of the item editing record corresponding to the trainee user according to the pre-item editing information and the post-item editing information of each piece of registered item information.

Next, a plurality of editing attribute units corresponding to the item editing record corresponding to the student user may be acquired, and a target editing attribute unit of the plurality of editing attribute units may be counted, where an editing character code feature exists in the target editing attribute unit.

Therefore, whether the associated editing items exist between two adjacent target editing attribute units can be judged, if yes, the quantity of the associated editing items is counted, when the quantity does not exceed a set numerical value, the item editing behavior set is implanted into each user registration data information, when the item editing behavior set implanted into each user registration data information is updated, the updated item editing behavior set is obtained, and the editing behavior characteristics and the deviation information of the registered item information corresponding to each acquired updated item editing behavior set are counted.

And then, determining the editing weight of each updated project editing behavior set according to the editing behavior characteristic corresponding to each updated project editing behavior set and the deviation information of the registered project information, correcting the updated project editing behavior set acquired in real time according to the editing weight to obtain a student user block editing sequence, extracting the learning behavior style data in the registered data information of each user according to the sequence characteristic in the student user block editing sequence, and determining the preset configuration attribute information of the student user according to the extracted learning behavior style data.

In a possible example, in the process of performing data feature identification on the trainee user according to each matching position to obtain the corresponding model matching attribute, the embodiment may obtain the current behavior feature of the trainee user and locate the first model feature corresponding to each matching position from the current behavior feature, and then determine whether the first model feature corresponding to each matching position in the current behavior feature has a matched feature value with respect to the second model feature in the current behavior feature. Wherein the second model feature is a feature other than the first model feature in the current behavior feature.

If the first model feature corresponding to each matching position in the current behavior features has a matched feature value relative to the second model feature in the current behavior features, the first model feature corresponding to each matching position located in the current behavior features can be determined as an effective model feature of the current behavior features, otherwise, the first model feature corresponding to each matching position located in the current behavior features and the second model feature in the current behavior features are subjected to weighted summation, and the weighted summation result is determined as the effective model feature of the current behavior features.

On this basis, for each matching position, a first model matching command line in the running thread of the matching position implantation server 100 is extracted, partial features in the effective model features of the current behavior features are fused with the first model matching command line to obtain a second model matching command line, and then the first model matching command line and the second model matching command line are respectively run in the mirror image threads corresponding to the running threads to obtain a first running result and a second running result which respectively correspond to each other.

Therefore, whether the similarity of the first operation result and the second operation result reaches a preset threshold value or not can be judged, when the similarity of the first operation result and the second operation result reaches the preset threshold value, the matching position is started to operate the second model matching command line to obtain a third operation result corresponding to the second model matching command line, the feature classification information in the third operation result is extracted, the model matching attribute corresponding to the matching position is obtained according to the feature classification information, and when the similarity of the first operation result and the second operation result does not reach the preset threshold value, the step of fusing partial features in the effective model features of the current behavior features with the first model matching command line to obtain the second model matching command line is returned.

It is worth noting that, on the basis, the inventor further considers that each second target online learning recommendation model may have different emphasis recommendation manners, and in order to avoid excessive screening experience brought to the student user by too much recommendation, in this embodiment, a second online learning recommendation result obtained by information recommendation of each second target online learning recommendation model to the student user according to the style attribute weight parameter may be further obtained, and then, according to the recommended content item of each second online learning recommendation result, the recommended style feature of the recommended style corresponding to each second online learning recommendation result may be obtained, so that the recommended screening result may be obtained according to the feature screening range of the preset recommended style of the student user and the recommended style feature of the recommended style corresponding to the second online learning recommendation result.

It should be noted that the recommendation screening result may include a plurality of style feature sets corresponding to the recommendation style features located in the feature screening range of the preset recommendation style.

Then, the filtering characteristic content information of any one first filtering characteristic content with different style characteristics contained in the recommended filtering result can be obtained, the filtering characteristic content attribute of the first filtering characteristic content is determined according to the filtering characteristic content information of the first filtering characteristic content, the target learning scene corresponding to the first filtering characteristic content is determined based on the filtering characteristic content scene in the filtering characteristic content information of the first filtering characteristic content, the recommended course information matched with the filtering characteristic content attribute of the first filtering characteristic content is determined at the same time, and the recommended course matched with the recommended course information is selected, so that the target recommended course matched with the first filtering characteristic content can be selected in the recommended courses with the recommended course information according to the filtering characteristic content attribute of the first filtering characteristic content and the course labels of the recommended courses with the recommended course information in the target learning scene, wherein the target recommended course further needs to match a second screening feature content associated with the first screening feature content.

Then, the course label information of the first screening feature content included in the screening feature content information of the first screening feature content may be acquired, and the course label information of the second screening feature content included in the screening feature content information of the second screening feature content may be acquired, so that a corresponding third online learning recommendation result may be generated according to the course label information of the first screening feature content and the course label information of the second screening feature content.

Based on the third online learning recommendation result obtained by the design integration screening, excessive screening experience brought to the student user by excessively complicated recommendation can be avoided

Fig. 3 is a schematic diagram of functional modules of an online learning recommendation device 300 according to an embodiment of the present application, where the online learning recommendation device 300 may be divided into the functional modules according to the foregoing method embodiment. For example, the functional blocks may be divided for the respective functions, or two or more functions may be integrated into one processing block. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. It should be noted that, the division of the modules in the present application is schematic, and is only a logical function division, and there may be another division manner in actual implementation. For example, in the case of dividing each function module according to each function, the online learning recommendation device 300 shown in fig. 3 is only a schematic device diagram. The online learning recommendation apparatus 300 may include an extraction module 310, a calculation module 320, a first recommendation module 330, and a second recommendation module 340, and the functions of the functional modules of the online learning recommendation apparatus 300 are described in detail below.

The extracting module 310 is configured to obtain learning behavior data corresponding to the learner user from each online learning terminal 200, and extract learning behavior style data corresponding to the learning behavior data, where the learning behavior data is obtained by the server 100 according to online learning mode information of the online learning terminal 200 corresponding to the learner user and an online learning interaction manner between the online learning terminal 200 and the server 100.

The calculating module 320 is configured to obtain a learning style weight parameter of the online learning recommendation model of each trainee user and a recommendation weight corresponding to the recommendation policy of each online learning recommendation model under the corresponding learning style weight parameter, and calculate an adaptation degree between the learning behavior data and each online learning recommendation model according to the style weight corresponding to the style data in the learning behavior style data, the learning style weight parameter of each online learning recommendation model and the recommendation weight corresponding to the recommendation policy under the corresponding learning style weight parameter.

The first recommending module 330 is configured to, when a target adaptation degree reaching a preset adaptation degree threshold exists in all the determined adaptation degrees, associate the student user with a first target online learning recommending model corresponding to the target adaptation degree, so that the first target online learning recommending model performs information recommendation on the student user to obtain a first online learning recommending result.

The second recommending module 340 is configured to, when there is no target adaptation degree reaching a preset adaptation degree threshold in all the determined adaptation degrees, determine a recommending level of each online learning recommending model, perform data feature recognition on learning behavior data of the learner user through each corresponding online learning recommending model according to a magnitude sequence of the recommending levels to obtain a plurality of model matching attributes corresponding to the learner user and a model attribute node corresponding to the recommending level of each online learning recommending model in each model matching attribute, determine a style attribute weight parameter corresponding to each model matching attribute from the learning behavior style data corresponding to the learner user according to each model matching attribute, and associate the style attribute weight parameter corresponding to each model matching attribute to a second target online learning recommending model corresponding to the model attribute node loaded in each model matching attribute, and performing information recommendation on the student users through each second target online learning recommendation model according to the style attribute weight parameters to obtain a second online learning recommendation result.

Further, fig. 4 is a schematic structural diagram of a server 100 for executing the online learning recommendation method according to an embodiment of the present application. As shown in FIG. 4, the server 100 may include a network interface 110, a machine-readable storage medium 120, a processor 130, and a bus 140. The processor 130 may be one or more, and one processor 130 is illustrated in fig. 4 as an example. The network interface 110, the machine-readable storage medium 120, and the processor 130 may be connected by a bus 140 or otherwise, as exemplified by the connection by the bus 140 in fig. 4.

The machine-readable storage medium 120 is a computer-readable storage medium, and can be used to store software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the online learning recommendation method in the embodiment of the present application (for example, the extraction module 310, the calculation module 320, the first recommendation module 330, and the second recommendation module 340 of the online learning recommendation apparatus 300 shown in fig. 3). The processor 130 executes various functional applications and data processing of the terminal device by detecting software programs, instructions and modules stored in the machine-readable storage medium 120, that is, the above online learning recommendation method is implemented, and details are not repeated herein.

The machine-readable storage medium 120 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like.

The processor 130 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method embodiments may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 130.

The server 100 can perform information interaction with other devices (such as the online learning terminal 200) through the network interface 110. Network interface 110 may be a circuit, bus, transceiver, or any other device that may be used to exchange information. Processor 130 may send and receive information using network interface 110.

In the above embodiments, the implementation may be wholly or partially implemented by software, hardware, firmware, or any pair thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the application to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device.

Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

It will be apparent to those skilled in the art that various changes and modifications may be made in the embodiments of the present application without departing from the spirit and scope of the application. Thus, to the extent that such expressions and modifications of the embodiments of the application fall within the scope of the claims and their equivalents, the application is intended to embrace such alterations and modifications.

Claims

1. The online learning recommendation method is applied to a server, wherein the server is in communication connection with a plurality of online learning terminals, and the method comprises the following steps:

2. The online learning recommendation method according to claim 1, wherein the step of calculating the degree of adaptation between the learning behavior data and each online learning recommendation model according to the style weight corresponding to the style data in the learning behavior style data, the learning style weight parameter of each online learning recommendation model and the recommendation weight corresponding to the learning style weight parameter in the recommendation policy under the corresponding learning style weight parameter comprises:

3. The online learning recommendation method according to claim 1, wherein the step of determining the recommendation level of each online learning recommendation model with respect to the trainee user comprises:

4. The online learning recommendation method according to claim 1, wherein the step of performing data feature recognition on the trainee user according to the recommendation level order to obtain a plurality of model matching attributes corresponding to the trainee user and a model attribute node loaded in each model matching attribute and corresponding to the recommendation level of each online learning recommendation model comprises:

5. The online learning recommendation method according to claim 4, wherein the step of determining the preset configuration attribute information of the trainee user comprises:

6. The online learning recommendation method according to claim 1, wherein the step of performing data feature recognition on the trainee user according to each matching position to obtain the corresponding model matching attribute comprises:

7. The online learning recommendation method according to any one of claims 1-6, characterized in that the method further comprises:

8. An online learning recommendation device is applied to a server, wherein the server is in communication connection with a plurality of online learning terminals, and the device comprises:

9. The online learning system is characterized by comprising a server and a plurality of online learning terminals in communication connection with the server;

10. A server, characterized in that the server comprises a processor, a machine-readable storage medium, and a network interface, the machine-readable storage medium, the network interface, and the processor are connected through a bus system, the network interface is used for being connected with at least one online learning terminal in a communication manner, the machine-readable storage medium is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the machine-readable storage medium to execute the online learning recommendation method according to any one of claims 1 to 7.