CN116662527A - Method for generating learning resources and related products - Google Patents

Abstract

Embodiments of the present invention provide a method for generating learning resources and related products. Wherein the method comprises the following steps: acquiring learning interaction information generated by a user in a learning process; analyzing the learning interaction information by using the trained neural network model to obtain behavior characteristics and learning content characteristics of the user; and generating learning resources associated with the user learning state according to the behavior characteristics and the learning content characteristics. According to the technical scheme, the actual learning states such as the user learning capacity and the actual learning requirements can be determined by utilizing the behavior characteristics and the learning content characteristics of the user, and learning resources related to the user learning states can be dynamically generated. Therefore, the learning resources of the user can be dynamically changed along with the learning state change of the user, so that the dynamic learning effect of thousands of people and thousands of faces is achieved, and the actual requirements of the user are met.

Description

Method for generating learning resources and related products

technical field

Embodiments of the present invention relate to the technical field of information processing, and more specifically, embodiments of the present invention relate to a method for generating learning resources, an electronic device and a computer-readable storage medium for executing the aforementioned method.

Background technique

This section is intended to provide a background or context for implementations of the invention that are recited in the claims. The descriptions herein may include concepts that could be explored, but not necessarily concepts that have been previously thought of or explored. Therefore, unless otherwise indicated herein, what is described in this section is not prior art to the description and claims in this application and is not admitted to be prior art by inclusion in this section.

How to effectively read and learn has attracted more and more attention from parents. Usually, parents will purchase a reading learning plan to help their children learn to read. At present, the mainstream reading and learning resource packages are all graded based on the user's age, that is, learning resources with a fixed theme or scope are output by inputting the user's age. This kind of learning resources obtained through content classification, its content is fixed, and finally achieves the same learning effect for everyone.

However, in practical applications, users' learning ability and needs are dynamically changing, and learning resources with fixed content obviously cannot meet users' learning needs.

Contents of the invention

Known learning resources with fixed content are not ideal for user learning assistance, which is a very annoying process.

Therefore, there is a great need for an improved solution for generating learning resources, which can dynamically generate learning resources associated with the user's learning status to meet the actual needs of users.

In this context, the embodiments of the present invention are expected to provide a method for generating learning resources and related products.

In the first aspect of the embodiments of the present invention, a method for generating learning resources is proposed, including: acquiring learning interaction information generated by users during the learning process; Information is analyzed to obtain the user's behavioral characteristics and learning content characteristics; and learning resources associated with the user's learning status are generated according to the behavioral characteristics and the learning content characteristics.

In an embodiment of the present invention, wherein the learning interaction information includes user behavior data and learning content data, analyzing the learning interaction information using a trained neural network model includes: separately analyzing the user behavior data based on the neural network model The behavior data and the learning content data are analyzed to obtain the behavior characteristics and the learning content characteristics.

In another embodiment of the present invention, wherein the neural network model includes a first network model and a second network model, respectively analyzing the user behavior data and the learning content data based on the neural network model includes: The first network model performs basic text analysis on the user behavior data and the learning content data; and based on the second network model, performs feature analysis on the output result of the first network model to obtain the behavior characteristics and the learning content features.

In yet another embodiment of the present invention, the first network model and the second network model are obtained through network model training based on the artificial intelligence generated content AIGC model.

In yet another embodiment of the present invention, generating learning resources associated with the user's learning state according to the behavior characteristics and the learning content characteristics includes: determining candidate content according to the behavior characteristics and the learning content characteristics; Screen out the learning resources associated with the user's learning status from the above candidate content.

In an embodiment of the present invention, determining the candidate content according to the behavior feature and the learning content feature includes: matching the behavior feature and the learning content feature in a predetermined database to obtain the candidate content.

In another embodiment of the present invention, screening the learning resources associated with the user's learning status from the candidate content includes: acquiring the user's forward learning record; Preliminary screening is performed to filter out the content that has been learned in the candidate content; and the learning resources are selected from the candidate content after the preliminary screening.

In yet another embodiment of the present invention, selecting the learning resources from the candidate contents after preliminary screening includes: acquiring the contents whose correlation degree with the completed learning content satisfies a predetermined threshold among the candidate contents; The weights are used to rank the obtained contents; and according to the ranking of the candidate contents, at least one candidate content is selected as the learning resource.

In the second aspect of the embodiments of the present invention, there is provided an electronic device, including: a processor; and a memory storing computer instructions for generating learning resources, when the computer instructions are executed by the processor, The electronic device is made to execute the methods described in the foregoing and following embodiments.

In the third aspect of the embodiments of the present invention, a computer-readable storage medium is provided, which contains program instructions for generating learning resources. When the program instructions are executed by a processor, the implementation of the above and below multiple The method described in the examples.

According to the method for generating learning resources and related products in the embodiments of the present invention, the learning resources can be dynamically generated according to the behavior characteristics and learning content characteristics analyzed from the learning interaction information. It can be seen that the solution of the present invention can use the user's behavior characteristics and learning content characteristics to determine the actual learning status of the user, such as the user's learning ability and actual learning needs, and dynamically generate learning resources associated with the user's learning status. In this way, the user's learning resources can be dynamically changed following the change of the user's learning status, thereby achieving a dynamic learning effect for thousands of people and meeting the actual needs of the user.

In addition, in some embodiments of the present invention, the trained neural network model may include a first network model that supports basic text analysis and a second network model that supports feature analysis, so that basic big data analysis can be performed based on the first network model In order to improve the computing efficiency of the entire network model, and superimpose the second network model for further detailed analysis to improve the computing accuracy of the entire network model. Thus, it can be ensured that the entire neural network model can have both operational efficiency and precision in the process of analyzing and learning interactive information.

Description of drawings

The above and other objects, features and advantages of exemplary embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are shown by way of illustration and not limitation, in which:

Figure 1 schematically illustrates a block diagram of an exemplary computing system 100 suitable for implementing embodiments of the present invention;

FIG. 2 schematically shows a schematic flowchart of a method for generating learning resources according to an embodiment of the present invention;

Fig. 3 schematically shows a schematic flowchart of a method for generating learning resources according to another embodiment of the present invention;

Fig. 4 schematically shows a schematic flowchart of a method for generating learning resources according to yet another embodiment of the present invention;

Fig. 5 schematically shows a schematic diagram of the training process of the AIGC model according to an embodiment of the present invention;

6 schematically shows a schematic diagram of the process of analyzing user behavior characteristics and learning content characteristics based on the AIGC model according to an embodiment of the present invention;

Fig. 7 schematically shows a schematic diagram of a process of adaptively generating learning resources according to an embodiment of the present invention;

FIG. 8 schematically shows a schematic diagram of a matching calculation process of user behavior features and learning content features according to an embodiment of the present invention; and

Fig. 9 schematically shows a schematic structural diagram of an electronic device according to an embodiment of the present invention.

In the drawings, the same or corresponding reference numerals denote the same or corresponding parts.

Detailed ways

The principle and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are given only to enable those skilled in the art to better understand and implement the present invention, rather than to limit the scope of the present invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

Figure 1 shows a block diagram of an exemplary computing system 100 suitable for implementing embodiments of the present invention. As shown in FIG. 1 , the computing system 100 may include: a central processing unit (CPU) 101, a random access memory (RAM) 102, a read only memory (ROM) 103, a system bus 104, a hard disk controller 105, and a keyboard controller 106 , serial interface controller 107, parallel interface controller 108, display controller 109, hard disk 110, keyboard 111, serial peripheral 112, parallel peripheral 113 and display 114. Among these devices, coupled to the system bus 104 are a CPU 101 , a RAM 102 , a ROM 103 , a hard disk controller 105 , a keyboard controller 106 , a serial controller 107 , a parallel controller 108 and a display controller 109 . Hard disk 110 is coupled with hard disk controller 105, keyboard 111 is coupled with keyboard controller 106, serial peripheral device 112 is coupled with serial interface controller 107, parallel peripheral device 113 is coupled with parallel interface controller 108, and display 114 is coupled with display control Device 109 is coupled. It should be understood that the structural block diagram shown in FIG. 1 is only for the purpose of illustration, rather than limiting the scope of the present invention. In some cases, some equipment can be added or subtracted on a case-by-case basis.

Those skilled in the art know that the embodiments of the present invention can be implemented as a system, method or computer program product. Therefore, the present disclosure can be embodied in the form of complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as "circuit", " module", "unit" or "system". Furthermore, in some embodiments, the present invention can also be implemented in the form of a computer program product embodied in one or more computer-readable media having computer-readable program code embodied therein.

Any combination of one or more computer readable medium(s) may be utilized. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (non-exhaustive examples) of computer-readable storage media may include, for example, an electrical connection with one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read only memory (ROM) , erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

A computer readable signal medium may include a data signal carrying computer readable program code in baseband or as part of a carrier wave. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. .

Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out the operations of the present invention may be written in one or more programming languages, or combinations thereof, including object-oriented programming languages—such as Java, Smalltalk, C++, and conventional Procedural Programming Language - such as "C" or a similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (such as through the Internet using an Internet service provider). connect).

Embodiments of the present invention will be described below with reference to flowcharts of methods and block diagrams of devices (or systems) in embodiments of the present invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations 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, or other programmable data processing apparatus to produce a machine, and these computer program instructions are executed by the computer or other programmable data processing apparatus to produce a flow diagram of the implementation and/or means for the functions/operations specified in the blocks in the block diagrams.

These computer program instructions can also be stored in a computer-readable medium that can cause a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable medium can generate a program including implementation flowcharts and and/or the product of the instruction device for the function/operation specified in the box in the block diagram.

It is also possible to load computer program instructions onto a computer, other programmable data processing apparatus, or other equipment, so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other equipment to produce a computer-implemented process, thereby Instructions that enable execution on a computer or other programmable device provide a process for implementing the functions/operations specified in the flowcharts and/or blocks in the block diagrams.

According to the embodiments of the present invention, a method for generating learning resources and related products thereof are proposed. In addition, any number of elements in the drawings is used for illustration rather than limitation, and any designation is only for distinction and does not have any limiting meaning.

The principle and spirit of the present invention will be explained in detail below with reference to several representative embodiments of the present invention.

Summary of the invention

The inventors of the present invention have found that the existing learning resources using fixed content are not effective in assisting users in learning. Specifically, the current mainstream reading learning solutions mainly provide fixed reading learning resource packages, and only support users to select specific learning resources for fixed learning. For example, content may be pre-graded to obtain resource pack 1, resource pack 2, resource pack 3, and so on. Wherein, the theme and content of each resource pack are fixed. In the specific use process, the user selects the resource package corresponding to the grade according to his actual age, so that the user can complete the learning of fixed content. As a result, the resource packs that different users of the same age group can choose are fixed and the content is exactly the same, and finally achieve the same learning effect for everyone. However, each user's learning ability and needs are dynamically changing during the learning process, and the fixed resource package obviously cannot meet the actual learning needs.

In related technologies, on the basis of content ratings, a solution for supporting user-defined learning plans may be added. However, the customized dimension mainly focuses on the management of learning time, for example, customized adjustments for the length of learning and frequency of learning. This custom adjustment of learning frequency or duration only involves a change in learning time, and does not change the content of the specific resource package, which still cannot meet the dynamic learning needs of users.

Based on this, the inventor found through research that the actual learning status such as the user's learning ability and actual learning needs can be determined by using the user's behavior characteristics and learning content characteristics, and learning resources associated with the user's learning status can be dynamically generated. In this way, the user's learning resources can be dynamically changed following the change of the user's learning status, so as to meet the actual learning needs of the user.

After introducing the basic principles of the present invention, various non-limiting embodiments of the present invention are described in detail below.

exemplary method

A method for generating learning resources according to an exemplary embodiment of the present invention is described below with reference to FIG. 2 . It should be noted that the embodiments of the present invention can be applied to any applicable scene.

Fig. 2 schematically shows a flowchart of a method 200 for generating learning resources according to an embodiment of the present invention.

As shown in FIG. 2 , at step S201 , learning interaction information generated by the user during the learning process may be acquired. The learning interaction information here can be understood as information generated by users in various learning scenarios, for example, it can include user age, user school age, user learning time, user learning content, content theme, content difficulty, feedback information from test exercises, etc. data. In a specific application, the learning interaction information may include information in various forms such as pictures, videos, texts, etc., and the learning interaction information may be acquired in various ways. For example, in some embodiments, an information input interface may be displayed, and the user may input the above-mentioned learning interaction information on the interface. In some other embodiments, the online learning system may provide users with learning scenarios, and collect various data generated by the users during the learning process in the learning scenarios. In addition, various intelligent learning terminals (such as dictionary pens, intelligent learning desk lamps, etc.) used by the user can also be linked to obtain the above-mentioned learning interaction information from these intelligent learning terminals. It should be noted that the detailed description of the learning interaction information here is only an exemplary illustration, and the solution of the present invention is not limited thereto, and may be specifically adjusted in combination with actual application scenarios.

Next, at step S202, the trained neural network model may be used to analyze the learning interaction information to obtain user behavior characteristics and learning content characteristics. The neural network model here is pre-trained. After the aforementioned learning interaction information is obtained, the learning interaction information can be input into the neural network model for analysis, so as to analyze the learning interaction information of the user during the learning process. Behavioral characteristics and learning content characteristics. For example, in some implementation scenarios, the user interaction information may include information such as the user's age, reading hobbies, reading ability, reading history, grade, type, and difficulty of the content itself. The neural network model can be used to analyze the user's age, reading hobbies, reading ability, reading records and other behavioral data, and get the behavioral characteristics of grade X users who like to read detective-themed books and have strong independent reading ability. The neural network model can also be used to analyze the content data such as the grade, type, and difficulty of the content itself, and obtain the content characteristics such as subject enlightenment, history, and plant-themed content that are suitable for reading by X-grade users. It should be noted that the detailed description of the user's behavior characteristics and learning content characteristics here is only an exemplary description. In a specific application, learning interaction information changes dynamically, and the analyzed behavior characteristics and learning content characteristics also change dynamically.

Finally, at step S203, learning resources associated with the user's learning status can be generated according to the aforementioned behavioral characteristics and learning content characteristics.

Wherein, the aforementioned behavior characteristics can represent the user's learning ability, and the learning content characteristics can be used to represent the user's demand for learning content. Thus, the user's actual learning state can be determined through the behavioral characteristics and learning content characteristics, and learning resources associated with the user's learning state can be dynamically generated.

It can be seen that the above solution of the present invention can realize the dynamic change of the user's learning resources following the change of the user's learning state, thereby achieving a dynamic learning effect with thousands of people and faces, and meeting the actual needs of the user.

Fig. 3 schematically shows a flowchart of a method 300 for generating learning resources according to another embodiment of the present invention. It can be understood that the method 300 is a further definition and/or extension of the method 200 in FIG. 2 . Therefore, the relevant detailed description above in conjunction with FIG. 2 is also applicable to the following.

As shown in FIG. 3 , at step S301 , user behavior data and learning content data generated by the user during the learning process can be acquired. In this embodiment, learning interaction information can be roughly divided into user behavior data and learning content data. The user behavior data may include but not limited to the user's school age, age, learning time, learning records and other data that can reflect the user's learning ability. The learning content data may include, but not limited to, the grade, type, difficulty, etc. of the content, which can reflect the information of the content itself. In a specific application, the above-mentioned user behavior data and learning content data can be obtained through various methods such as active uploading by the user, automatic acquisition in the background, or linkage with intelligent learning terminals such as dictionary pens and desk lamps used in the user's learning process. It should be noted that the detailed description of the user behavior data and learning content data here is only an example, and the solution of the present invention is not limited thereto.

Next, at step S302, basic text analysis may be performed on the above user behavior data and learning content data by using the first network model. In some embodiments, the trained neural network model may specifically include a first network model and a second network model. Among them, the first network model can be used to solve the computing power problem of the entire neural network model, and it is specifically configured to perform basic text analysis on user behavior data and learning content data. For example, basic text analysis operations such as data cleaning, data format conversion, and data normalization processing can be performed.

At step S303, the second network model may also be used to perform feature analysis on the output result of the first network model to obtain behavioral features and learning content features. The second network model here can be used to solve the calculation accuracy of the entire neural network model, and its specific configuration is to perform feature extraction on the output results obtained after analysis by the first network model, and the first network model shares large-scale basic data Analysis, the second network model is responsible for accurate feature analysis.

It should be noted that the deep neural network model responsible for analyzing learning interaction information (including user behavior data and learning content data) here adopts a dual network model (namely, the first network model and the second network model). In some embodiments, the first network model and the second network model may be obtained through network model training based on an artificial intelligence generated content (AI Generated Content, AIGC for short) model. The AIGC model can be understood as a large language model similar to GPT3.0 and above, and this type of model has strong text analysis and content generation capabilities. Use the trained first network model for basic big data analysis to improve the operational efficiency of the entire network model, and superimpose the trained second network model for further fine analysis to improve the operational accuracy of the entire network model. Thus, it can be ensured that the entire neural network model can have both operational efficiency and precision in the process of analyzing and learning interactive information.

The above-mentioned specific analysis process of data using the double network model is only an exemplary description. For example, in practical applications, the deep neural network model can also adopt a single network model architecture to implement the above basic text parsing operations and feature parsing operations based on the single network model architecture.

After the behavioral features and learning content features are obtained, at step S304, candidate content can be determined according to the behavioral features and learning content features. For example, in some embodiments, the aforementioned behavioral features and learning content features can be matched in a predetermined database to obtain candidate content. It should be noted that the description of the process of determining the candidate content here is only an exemplary description, and the solution of the present invention is not limited thereto. For example, the obtained behavioral features and learning content features can also be uploaded to the cloud or server side for matching and comparison to obtain candidate content.

Finally, at step S305, the learning resources associated with the user's learning status can be screened out from the aforementioned candidate content. Specifically, the user's forward learning record can be acquired. For example. The forward learning record can be obtained in various ways such as uploading actively by the user or linking with other intelligent learning terminals. Wherein, the forward learning record may include content that the user has completed learning or content that the user is not interested in. Then, the candidate content can be preliminarily screened based on the forward learning record, so as to filter out the learned content or the content that the user is not interested in among the candidate content. Then, the learning resources are selected from the candidate content after the preliminary screening.

In some embodiments, specifically, among the candidate content, the content whose degree of relevance to the learned content satisfies a predetermined threshold may be acquired. Then, the acquired content may be sorted according to content weights, and at least one candidate content is selected as a learning resource according to the sorting of candidate content. Among them, the greater the content weight of the candidate content, the higher the degree of correlation with the user's learning status, and the higher the ranking, that is, the greater the probability of being selected as a learning resource. The predetermined threshold here can be set and adjusted according to actual application requirements. The content weight can be determined according to the artificial setting value and the system calculation assignment. Among them, the system calculation assignment is obtained according to the weighted sum of the calculation indicators of the content, and the calculation indicators involved (such as the click effect of the content on the platform, search efficiency, reading effect, etc.) can be adjusted and set according to actual needs. It should be noted that the detailed description of the learning resource screening process here is only an example, and the solution of the present invention is not limited thereto.

Therefore, by using the user's behavioral characteristics and learning content characteristics to determine the actual learning status of the user, such as the user's learning ability and actual learning needs, and dynamically generate learning resources associated with the user's learning status, the user's learning resources can follow the user's learning status. The dynamic changes have truly realized the dynamic learning effect of thousands of people and faces. In addition, with the help of the text analysis and content generation capabilities of the AIGC model, the efficiency of user and content matching is greatly improved, and the adaptability of learning resources is efficiently realized.

Fig. 4 schematically shows a flowchart of a method 400 for generating learning resources according to yet another embodiment of the present invention. It can be understood that the method 400 can be understood as a specific technical implementation of the method 200 or the method 300 . Therefore, the foregoing detailed descriptions in conjunction with FIG. 2 and FIG. 3 also apply to the following.

As shown in FIG. 4 , at step S401 , AIGC model training for educational scenarios can be performed. In some embodiments, based on the AIGC model training platform, data sets based on current business scenarios (such as education scenarios) can be provided for model training, so as to generate a model with sufficient understanding of the scenario business.

Fig. 5 exemplarily shows a feasible training method for the AIGC model in educational scenarios. As shown in FIG. 5 , at step S501 , data preparation can be performed. Specifically, the AIGC model of the education scene may include the AIGC basic large model (that is, the first network model above) and the AIGC scene small model (that is, the second network model above). Therefore, in the data preparation stage, training data and test data can be prepared separately for these two models.

At step S502, model selection is performed. Specifically, it is necessary to select an appropriate AIGC model as the basic model for training based on business requirements. In some embodiments, multiple selection dimensions such as data preprocessing, feature selection, machine learning algorithms, and evaluation methods need to be considered during the model selection process. For example, AIGC's basic large model needs to support basic text analysis, which requires strong data processing capabilities to solve computing power problems. In some implementation scenarios, language model GPT-4, language representation model (BERT model for short), powerfully optimized language representation model (RoBERTa model for short), etc. can be used as the basic model for training the AIGC basic large model. The small model of the AIGC scene needs to support feature analysis, which requires precise analysis capabilities to solve the accuracy problem. In some implementation scenarios, the Pegasus model (referred to as the PEGASUS model), the unified language model (referred to as the UniLM model), etc. can be used as the basic model for training the small model of the AIGC scenario. It should be noted that the description of the models that can be used by the basic model here is only an illustration, and the solution of the present invention is not limited thereto.

At step S503, data preprocessing may be performed. For example, training data can be preprocessed, specifically including operations such as data cleaning, data format conversion, and data normalization. At step S504, feature selection can be performed. Specifically, when performing feature selection on training data, it is necessary to select the features that are most helpful to the problem. For example, the user's school age can highlight the user's learning ability more than his age. At this time, the user's school age can be selected as the behavior feature. At step S505, model training can be performed. After preparing the training data, the basic model, etc., you can use the training data to train the AIGC model as the basic model, and continuously adjust the model parameters to make the model performance optimal. At step S506, model evaluation can be performed. Specifically, the aforementioned test data can be used to evaluate the trained model, and the model can be further optimized according to the evaluation result. At step S507, the application of the model can be implemented. Specifically, the trained model can be applied to practical problems (such as the extraction of behavioral features and learning content features in educational scenarios, etc.) for prediction and decision-making. At step S508, model iteration can be performed. Specifically, the model can be iterated and optimized according to the prediction results of the model in actual application to ensure the validity and reliability of the model. So far, the training of the AIGC model for educational scenarios has been completed.

In this embodiment, the AIGC model of the education scene includes the AIGC basic large model and the AIGC small model of the scene, so the training process needs to train these two models separately. Of course, if the AIGC basic large model itself can complete the basic text analysis well, it can also be trained only for the small model of the AIGC scene. In addition, in practical applications, the AIGC model for educational scenarios can also be a single AIGC model that combines basic text parsing and feature parsing. For the training of the single AIGC model, reference can be made to the above model training process, which will not be repeated here.

Returning to FIG. 4 , at step S402 , the trained AIGC model for educational scenarios can be used to analyze user behavior and learning content. Specifically, by inputting user behavior data and learning content data, the AIGC model of the trained educational scene is used for analysis to generate corresponding user behavior feature data and learning content feature data for the establishment and use of subsequent matching relationships.

Fig. 6 exemplarily shows a feasible way to analyze user behavior and learning content by using the trained AIGC model for educational scenarios. As shown in Figure 6, in terms of model selection, model superposition can be used, using the AIGC basic large model that supports basic text analysis, and superimposing the AIGC small scene model that supports feature analysis for use. In practical applications, the proportion of information processed can be adjusted by adjusting the degree of influence of each model in the AIGC model in educational scenarios.

Next, at step S403, learning resource matching can be performed. Specifically, matching calculations can be performed on the obtained behavioral features and learning content features in the database, so as to finally filter out the learning resources associated with the user's learning status. As shown in Figure 7, the user status changes dynamically during the user learning process, and learning resources related to different user status can be matched according to the behavior characteristics and learning content characteristics. For example, when the user status is 1, learning resource 1 to learning resource 4 can be matched. As the user status changes, when it becomes user status 2, learning resource 1 and learning resources 5-7 can be matched. When changing to user status 3, learning resources 8-10 can be matched. It should be noted that this is only an exemplary illustration of the user state change process and its associated learning resources, and the division of user states and the number of learning resources are not limited.

In some embodiments, the matching calculation process of behavioral features and learning content features may specifically involve processes such as screening content, filtering, sorting, and value selection. FIG. 8 exemplarily shows a possible matching calculation process of behavioral features and learning content features. As shown in FIG. 8, at step S801, content can be screened. Specifically, match the user features calculated and generated by the trained AIGC model with the learning content features in the database. For example, the database contains a variety of learning resources with tags, which can be based on user behavior characteristics (that is, user tags) such as the user's age, reading ability, and reading interest, as well as the content's suitable reading age, content theme, content adaptability, etc. Learn content features (that is, content labels) for matching to obtain candidate content. Next, at step S802, filtering may be performed. Specifically, the user's forward reading records can be combined to filter the content that the user has completed learning in the candidate content.

Then, at step S803, sorting can be performed. Specifically, according to the correlation index between the contents, the content with high correlation with the content that the user has currently completed learning can be preferentially obtained (such as content similar to the content theme or content adaptability level), and then according to the weight of the content. Sort. Wherein, the content weight may include two parts: human intervention (that is, the artificially set value mentioned above, for example, may account for 50%), and system calculation and assignment (for example, may account for 50%). In some embodiments, human intervention can be given by platform reading experts based on industry experience, and the value range is 0-5 points (the specific value range is not limited, and can be adjusted according to actual needs). The system calculation and assignment can be calculated based on the weighted calculation indicators of content on the platform, such as click effects, search effects, reading effects, payment conversion effects, etc., and the value range is 0-5 points (the specific value range is not limited, It can be adjusted according to actual needs). For example, system calculation assignment = click score * 0.2 + search score * 0.3 + reading score * 0.3 + paid conversion score * 0.2. Among them, the relevant calculation indicators and proportions in the formula are only illustrative descriptions, and can be adjusted according to actual business conditions.

Finally, at step S804, screening is performed to obtain values. Specifically, the candidate contents may be sorted according to the weight of the contents. For example, the candidate content can be sorted from high to low, and then combined with the number of content resources to be output (specifically, it can be adjusted according to the scene requirements), the value is selected from high to low. So far, the matching calculation of user behavior features and learning content features is completed.

Returning to Fig. 4, at step S404, the adaptive learning resources may be updated. For a continuous learning user, during the learning process, with the change of his own learning behavior and ability, he will adaptively obtain matching learning resources. Wherein, there may be one or more learning resources associated with each user state, and these learning resources may be displayed to the user sequentially, or simultaneously, for the user to select and learn at will. In addition, the specific types of learning resources are not limited here, for example, they may include reading content, exercise questions and other content suitable for user learning.

The solution of the present invention will be further explained below in conjunction with specific application scenarios.

Practical application scenario: For a grade 1 user, generate a learning resource to help him read and learn. Wherein, the learning materials may include but not limited to reading books, reading questions and the like.

In order to meet the needs of this application scenario, you can first collect training data and test data such as user reading behavior data and reading content data in the reading learning scenario for model training, so as to obtain an AIGC that can fully understand the user's reading learning ability and reading content. Model. Then, the user's age, reading hobbies, reading ability, reading records and other behavioral data can be input into the trained AIGC model for analysis, and the calculated behavioral characteristics. For example, first-grade children like to read detective-themed books and have strong independent reading ability and other behavioral characteristics. At the same time, the learning content data such as the grade, type, and difficulty of the content itself can be input into the trained AIGC model for analysis, and the characteristics of the learning content can be output. For example, the content suitable for grade 1 children is subject enlightenment, history, plant-themed content, content difficulty level and other learning content characteristics. It should be noted that the parsing process of behavioral features and learning content features is dynamically updated.

Then, the obtained user behavior characteristics can be matched with the learning content characteristics. For example: Grade 1 likes the theme of detectives, and has a strong reading ability, matching books with detective themes and reading ability requirements beyond grade 1, and matching reading comprehension questions (such as multiple-choice questions, etc.) that suit the user's ability. Finally, as the user's reading ability and needs change during the learning process, corresponding learning resources can be acquired adaptively. For example, although the user is in the first grade, according to the reading tracking for a period of time, many reading books and more difficult reading questions for higher grades are recommended to him. Similarly, if it is found that the user's reading effect is not good for a period of time, and the reading ability has not improved, it will continue to push more easy-to-learn learning content and simpler reading questions.

In this way, the results of AIGC model analysis are used for matching calculations between user information and content, and through dynamic matching, the effect of adaptively generating different learning resources according to different states of users is realized.

exemplary device

After introducing the method of the exemplary embodiment of the present invention, next, related products of the method for generating learning resources of the exemplary embodiment of the present invention will be described with reference to FIG. 9 .

Fig. 9 schematically shows a schematic block diagram of an electronic device 900 according to an embodiment of the present invention. As shown in FIG. 9 , an electronic device 900 may include a processor 901 and a memory 902 . The memory 902 stores computer instructions for generating learning resources. When the computer instructions are executed by the processor 901 , the electronic device 900 executes the method described above in conjunction with FIGS. 2 to 5 and 8 . For example, in some embodiments, the electronic device 900 can acquire learning interaction information, train a neural network model that supports the analysis of learning interaction information, analyze behavioral characteristics and learning content characteristics, and adaptively and dynamically generate learning resources associated with the user's learning status. etc. Based on this, the electronic device 900 can determine the actual learning status of the user such as the user's learning ability and actual learning needs by using the user's behavior characteristics and learning content characteristics, and dynamically generate learning resources associated with the user's learning status. It changes dynamically following the change of the user's learning status, and truly realizes the dynamic learning effect of thousands of people.

It should be noted that although in the above detailed description several means or sub-units of the apparatus are mentioned, such division is by no means mandatory. Actually, according to the embodiment of the present invention, the features and functions of two or more devices described above may be embodied in one device. Conversely, the features and functions of one device described above may be further divided to be embodied by a plurality of devices.

The use of the verbs "comprise", "comprise" and their conjugations mentioned in the application documents does not exclude the presence of elements or steps other than those stated in the application documents. The article "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.

Although the spirit and principles of the invention have been described with reference to a number of specific embodiments, it should be understood that the invention is not limited to the specific embodiments disclosed, nor does division of aspects imply that features in these aspects cannot be combined to achieve optimal performance. Benefit, this division is only for the convenience of expression. The present invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. The scope of the appended claims is to be accorded the broadest interpretation thereby encompassing all such modifications and equivalent structures and functions.

Claims (10)

1. A method for generating learning resources, comprising:

acquiring learning interaction information generated by a user in a learning process;

analyzing the learning interaction information by using the trained neural network model to obtain behavior characteristics and learning content characteristics of the user; and

And generating learning resources associated with the learning state of the user according to the behavior characteristics and the learning content characteristics.

2. The method of claim 1, wherein the learning interaction information includes user behavior data and learning content data, and wherein analyzing the learning interaction information using the trained neural network model comprises:

and analyzing the user behavior data and the learning content data based on a neural network model respectively to obtain the behavior characteristics and the learning content characteristics.

3. The method of claim 2, wherein the neural network model comprises a first network model and a second network model, and analyzing the user behavior data and the learning content data based on the neural network model comprises:

performing basic text parsing on the user behavior data and the learning content data based on the first network model; and

and carrying out feature analysis on the output result of the first network model based on the second network model so as to obtain the behavior feature and the learning content feature.

4. The method of claim 3, wherein the first network model and the second network model are trained based on an artificial intelligence generation content AIGC model.

5. The method of claim 1, wherein generating a learning resource associated with a user learning state from the behavioral characteristics and the learning content characteristics comprises:

determining candidate content according to the behavior characteristics and the learning content characteristics; and

learning resources associated with a user learning state are screened from the candidate content.

6. The method of claim 5, wherein determining candidate content from the behavioral characteristics and the learning content characteristics comprises:

and matching the behavior characteristic and the learning content characteristic in a preset database to obtain the candidate content.

7. The method of claim 5, wherein screening learning resources associated with a user learning state from the candidate content comprises:

acquiring a forward learning record of the user;

performing preliminary screening on the candidate content based on the forward learning record so as to filter the content which has completed learning in the candidate content; and

and screening the learning resources from the candidate content after the primary screening.

8. The method of claim 7, wherein screening the learning resources from the pre-screened candidate content comprises:

Acquiring content of which the correlation degree with the content which has completed learning in the candidate content meets a preset threshold value;

sorting the acquired content according to the content weight; and

and screening at least one candidate content as the learning resource according to the ranking of the candidate content.

9. An electronic device, comprising:

a processor; and

a memory storing computer instructions for generating a learning resource, which when executed by the processor, cause the electronic device to perform the method of any of claims 1-8.

10. A computer readable storage medium, characterized by containing program instructions for generating learning resources, which when executed by a processor, cause the method according to any of claims 1-8 to be implemented.