A method, apparatus, device, and storage medium for recommending exercises.

Abstract

This application provides a method, apparatus, device, and storage medium for recommending practice questions, including: acquiring historical answer records of X practice questions for a target object, and knowledge point information related to each of the X practice questions; constructing a feature representation of the target object based on the historical answer records and knowledge point information; inputting the feature representation into a cognitive diagnostic model, so that the cognitive diagnostic model outputs candidate practice questions; acquiring target parameters, which are weight parameters of the cognitive diagnostic model; constructing a multi-objective optimization problem and constraints for the multi-objective optimization problem based on the target parameters; and solving the multi-objective optimization problem based on at least two constraints, so as to obtain recommended practice questions from the candidate practice questions. The technical solution provided by this application can be applied to the fields of cloud computing or artificial intelligence, and can be used to flexibly recommend practice questions to the target object, thereby achieving wide applicability of the practice question recommendation method and meeting many application scenarios.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more particularly to a method, apparatus, device, and storage medium for recommending exercises. Background Technology

[0002] With the rapid development of online education, traditional educational practices are gradually transforming into digital education. A vast amount of educational resources are being shared on online education platforms for learners and educators to download and study. Tests and exercises for various subjects are widely used by learners and educators as important educational resources to reinforce knowledge learned in the classroom.

[0003] However, it is difficult for learners or educators to directly sift through a massive amount of exercises to find those that are truly suitable for learners. In traditional education, educators usually need to find relevant exercises from resources such as review materials, exercise sets, or past exam papers based on current exam requirements or knowledge point consolidation needs, and then select exercises that meet the requirements based on their personal experience.

[0004] Considering that manual test paper generation requires manually searching for questions, which is time-consuming and labor-intensive, automatic test paper generation technology has emerged. Current automatic test paper generation requires extracting questions from a question bank within the constraints of user-inputted parameters. For example, this can be achieved through random sampling or backtracking. The aforementioned automatic test paper generation technology defines test paper generation as an optimization problem under constrained combinations. However, since this combinatorial optimization is a non-deterministic polynomial (NP) problem, its optimization time efficiency is low, making it difficult to search for optimal solutions from massive question banks. Furthermore, the test papers generated based on these technologies are mostly general assessment questions, failing to meet the needs of targeted student assessment, and therefore cannot accurately reflect students' knowledge mastery and teaching quality. Summary of the Invention

[0005] This application provides a method, apparatus, device, and storage medium for recommending exercises to a target object in a flexible manner, thereby achieving wide applicability of the exercise recommendation method and meeting various application scenarios.

[0006] In view of this, this application provides a method for recommending practice questions, comprising: obtaining historical answer records of X practice questions for a target object, and knowledge point information related to each of the X practice questions, where X is a positive integer; constructing a feature representation of the target object based on the historical answer records and the knowledge point information; inputting the feature representation into a cognitive diagnostic model so that the cognitive diagnostic model outputs candidate practice questions; and obtaining target parameters, which are weight parameters of the cognitive diagnostic model, including the knowledge relevance between knowledge points related to the practice questions, the target object's proficiency in the knowledge points related to the practice questions, and the knowledge of the practice questions for the target object. The objective parameters are: difficulty level and knowledge point differentiation among the knowledge points involved in the exercises; a multi-objective optimization problem and its constraints are constructed based on these objective parameters. These constraints include at least the target object's proficiency with the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the recommended exercises' knowledge point difficulty reaching a first threshold and the knowledge relevance between the recommended exercises' knowledge points and the knowledge points involved in the X exercises reaching a second threshold. The multi-objective optimization problem is solved based on these at least two constraints to obtain the recommended exercises from the candidate exercises.

[0007] This application also provides a problem recommendation device, including: an acquisition module, used to acquire historical answer records of X problems of a target object, and knowledge point information involved in each of the X problems, where X is a positive integer;

[0008] The processing module is used to construct a feature representation of the target object based on the historical answer record and the knowledge point information; and input the feature representation into the cognitive diagnostic model so that the cognitive diagnostic model outputs candidate exercises.

[0009] The acquisition module is also used to acquire target parameters, which are the weight parameters of the cognitive diagnostic model. The weight parameters include the knowledge relevance between the knowledge points involved in the exercises, the knowledge proficiency of the target object in the knowledge points involved in the exercises, the difficulty of the exercises for the target object in the knowledge points, and the knowledge point differentiation between the knowledge points involved in the exercises.

[0010] The processing module is also used to construct a multi-objective optimization problem and constraints on the multi-objective optimization problem based on the target parameters. The constraints include at least the target object's knowledge proficiency in the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the difficulty of the knowledge points in the recommended exercises reaching a first threshold and the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaching a second threshold.

[0011] The output module is used to solve the multi-objective optimization problem based on the at least two constraints, so as to obtain the recommended problem from the candidate problems.

[0012] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is specifically used to train an initial cognitive diagnostic model based on the feature representation to obtain the cognitive diagnostic model, wherein the weight parameters of the initial cognitive diagnostic model include initial knowledge proficiency, initial knowledge relevance, initial knowledge point difficulty, and initial knowledge point discrimination.

[0013] Inputting this feature representation into the cognitive diagnostic model yields the predicted score for each exercise in the exercise bank;

[0014] The candidate exercise is output from the exercise bank based on the predicted score.

[0015] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is specifically used to establish the initial cognitive diagnostic model;

[0016] The hidden capability representation vector of the target object is obtained by passing this feature representation through the first layer network of the initial cognitive diagnostic model;

[0017] The hidden ability representation vector is passed through the second layer of the initial cognitive diagnostic model to obtain the initial knowledge proficiency.

[0018] The feature representation is used to obtain the initial knowledge relevance through the exercise knowledge point association matrix in the initial cognitive diagnostic model, the feature representation is used to obtain the initial knowledge point difficulty through the exercise difficulty training matrix in the initial cognitive diagnostic model, and the feature representation is used to obtain the initial knowledge point discrimination through the exercise discrimination training matrix in the initial cognitive diagnostic model.

[0019] Calculate the predicted response for each of the X exercises based on the initial knowledge proficiency, the initial knowledge relevance, the initial knowledge difficulty, and the initial knowledge discrimination.

[0020] The loss value of the initial cognitive diagnostic model is obtained using cross-entropy based on the predicted response and the actual responses of the X exercises.

[0021] The initial cognitive diagnostic model is obtained by adjusting the weight parameters of the cognitive diagnostic model in reverse based on the loss value. The weight parameters of the cognitive diagnostic model are the knowledge proficiency, the knowledge relevance, the knowledge difficulty, and the knowledge discrimination.

[0022] In one possible design, in another implementation of another aspect of the embodiments of this application, the output module is specifically used to solve the multi-objective optimization problem from the candidate problems using a dual-encoding and dual-population co-evolutionary algorithm, such that the problems that satisfy the at least two constraints are used as the recommended problems.

[0023] In one possible design, in another implementation of another aspect of the embodiments of this application, the output module is specifically used to construct a main population and an auxiliary population based on the candidate exercise. The main population includes initial population individuals oriented towards the direction of diversity evolution, and the auxiliary population includes initial population individuals oriented towards the direction of convergence evolution.

[0024] An auxiliary profile is obtained by performing optimization calculations on the parent individuals in the auxiliary population. The auxiliary profile includes the optimized individuals obtained from the optimization calculations on the auxiliary population.

[0025] Based on the guiding individuals in the auxiliary file and the parent individuals of the main population, the optimization calculation of the main population is performed to obtain the optimized individuals corresponding to the main population;

[0026] The recommended exercise is derived from the optimal individuals of the auxiliary population and the optimal individuals of the main population.

[0027] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is specifically used to generate exercise combinations from the candidate exercises using binary encoding to obtain the initial population individuals of the main population, and to generate exercise combinations from the knowledge points involved in the candidate exercises using knowledge point integer encoding to obtain the initial population individuals of the auxiliary population.

[0028] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is specifically used to select any individual from the auxiliary file as the guiding individual, and obtain the knowledge point integer code of the guiding individual;

[0029] Take any two individuals from the main population as the parent individuals, and obtain the knowledge point integer codes of the parent individuals;

[0030] The difference in the number of exercises corresponding to the knowledge points between the guided individual and the parent individual is calculated by comparing the integer codes of the knowledge points of the guided individual with those of the parent individual.

[0031] Calculate the flipping probability of each knowledge point in the parent individual based on the quantitative difference, and flip each knowledge point in the parent individual according to the flipping probability to obtain the crossover individual;

[0032] The flipping flag of each knowledge point in the crossover individual is determined based on the flipping probability and the first random value. The flipping flag is used to flip each knowledge point in the crossover individual to obtain a mutant individual, and the mutant individual is used as the offspring individual.

[0033] Repeat the above process until the optimization calculation is completed and the optimized individuals corresponding to the main population are output.

[0034] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is further configured to generate a second random value;

[0035] The second random value is compared with the proportion parameter to obtain the comparison result. The proportion parameter is used to indicate the probability value of the main population receiving the guidance strategy of the auxiliary population.

[0036] When the comparison result indicates that the second random value is greater than or equal to the ratio parameter, the operation of performing optimization calculations on the main population based on the guide individuals in the auxiliary file and the parent individuals of the main population to obtain the optimized individuals corresponding to the main population is triggered.

[0037] When the comparison result indicates that the second random number is less than the ratio parameter, the optimization calculation of the main population is performed based on the parent individuals in the main population to obtain the optimized individuals.

[0038] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is specifically used to arbitrarily select two parent individuals from the main population and flip the target knowledge points in the parent individuals to obtain crossover individuals.

[0039] Each knowledge point in the crossover individual is flipped according to the mutation probability to obtain a mutated individual, and this mutated individual is used as the offspring individual;

[0040] Repeat the above process until the optimization calculation is completed and the optimized individuals corresponding to the main population are output.

[0041] In one possible design, in another implementation of another aspect of the embodiments of this application, the processing module is specifically used to construct a first objective and a first constraint in the multi-objective optimization problem based on the knowledge proficiency in the target parameter. The first objective is that the knowledge point difficulty of the recommended exercise reaches a first threshold, and the first constraint is the knowledge proficiency constraint of the target object on the knowledge points involved in the recommended exercise.

[0042] Based on the knowledge relevance in the objective parameter, a second objective and a second constraint are constructed in the multi-objective optimization problem. The second objective is that the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaches a second threshold. The second constraint is a constraint on the number of knowledge points involved in the recommended exercises.

[0043] In one possible design, in another implementation of another aspect of the embodiments of this application, the constraint further includes a constraint on the number of recommended exercises and a constraint that the recommended exercises are unanswered questions for the target object.

[0044] In one possible design, in another implementation of another aspect of the embodiments of this application, the objectives of the multi-objective optimization problem further include the number of knowledge points involved in the recommended exercises reaching a third threshold and the number of recommended exercises reaching a fourth threshold.

[0045] This application also provides a computer device, including: a memory, a processor, and a bus system;

[0046] The memory is used to store programs;

[0047] The processor is used to execute programs in memory, and the processor is used to execute the methods mentioned above according to the instructions in the program code;

[0048] Bus systems are used to connect memory and processor to enable communication between them.

[0049] Another aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described above.

[0050] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the above aspects.

[0051] As can be seen from the above technical solutions, the embodiments of this application have the following advantages: the knowledge mastery of the target object is obtained through a cognitive diagnostic model, and then a corresponding multi-objective optimization problem and its constraints are constructed based on the target object's knowledge mastery. Finally, the multi-objective optimization problem is solved to obtain recommended exercises. Since the multi-objective optimization problem can be flexibly adjusted according to the actual situation, exercises can be flexibly recommended to the target object, thereby achieving the wide applicability of the exercise recommendation method and meeting many application scenarios. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of a cognitive diagnosis process in an embodiment of this application;

[0053] Figure 2 This is a flowchart illustrating the adaptive question grouping process in an embodiment of this application.

[0054] Figure 3 This is a schematic diagram of the architecture of an application system for the exercise recommendation scheme in this application embodiment;

[0055] Figure 4 This is a schematic diagram of one embodiment of the exercise recommendation method in this application.

[0056] Figure 5 This is a schematic diagram of the structure of a cognitive diagnostic model in an embodiment of this application;

[0057] Figure 6 This is a schematic diagram illustrating the encoding of individuals in the main population and auxiliary population in the embodiments of this application;

[0058] Figure 7 This is a schematic diagram illustrating the process of collaborative optimization between the main population and the auxiliary population in an embodiment of this application;

[0059] Figure 8 This is a schematic diagram illustrating the process of guided optimization between the main population and the auxiliary population in an embodiment of this application;

[0060] Figure 9a This is a schematic diagram of the exercise generation interface in an embodiment of this application;

[0061] Figure 9b This is another schematic diagram of the exercise generation interface in the embodiments of this application;

[0062] Figure 10 This is a schematic diagram of the exercise generation interface in an embodiment of this application;

[0063] Figure 11 This is another schematic diagram of the exercise generation interface in the embodiments of this application;

[0064] Figure 12 This is a schematic diagram of one embodiment of the exercise recommendation device in this application;

[0065] Figure 13 This is a schematic diagram of another embodiment of the exercise recommendation device in this application;

[0066] Figure 14 This is a schematic diagram of another embodiment of the exercise recommendation device in this application. Detailed Implementation

[0067] This application provides a method, apparatus, device, and storage medium for recommending exercises to a target object in a flexible manner, thereby achieving wide applicability of the exercise recommendation method and meeting various application scenarios.

[0068] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0069] With the rapid development of online education, traditional educational practices are gradually transforming into digital education. A vast amount of educational resources are shared on online education platforms for learners and educators to download and study. Test questions for various subjects are widely used by learners and educators as important educational resources to reinforce classroom learning. However, it is difficult for learners and educators to directly sift through the massive amount of questions to find those truly suitable for their learners. In traditional education, educators typically need to search for relevant questions from review materials, question sets, or past exam papers based on current exam requirements or knowledge point reinforcement needs, and then select suitable questions based on personal experience. Considering that manually creating test papers requires manually searching for questions, which is time-consuming and labor-intensive, automatic test paper creation technology has emerged. Current automatic test paper creation requires extracting questions from a question bank within the constraints of user-input parameters, such as through random sampling strategies or backtracking methods. The aforementioned automatic test paper generation technology defines test paper generation as an optimization problem under constrained combination. However, since this combination optimization is a non-deterministic polynomial (NP) problem, the optimization time efficiency is low, and it is difficult to search for the optimal solution from a massive question bank. Furthermore, the test papers generated based on the above technology are mostly general assessment questions, which cannot meet the needs of targeted assessment of students. Therefore, they cannot well reflect students' knowledge mastery and teaching quality.

[0070] To address the aforementioned issues, this application provides the following technical solution: Obtaining historical answer records for X exercises from the target object, and the knowledge point information involved in each of the X exercises, where X is a positive integer; constructing a feature representation of the target object based on the historical answer records and the knowledge point information; inputting the feature representation into a cognitive diagnostic model to output candidate exercises; obtaining target parameters, which are weight parameters of the cognitive diagnostic model, including the knowledge relevance between the knowledge points involved in the exercises, the target object's proficiency in the knowledge points involved in the exercises, and the difficulty of the exercises for the target object's knowledge points. The objective parameters are: the degree of knowledge proficiency of the target object in the recommended exercises and the degree of knowledge differentiation among the knowledge points involved in the exercises; a multi-objective optimization problem and its constraints are constructed based on the objective parameters. The constraints include at least the knowledge proficiency of the target object in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the knowledge difficulty of the recommended exercises reaching a first threshold and the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaching a second threshold. The multi-objective optimization problem is solved based on the at least two constraints so that the recommended exercises are obtained from the candidate exercises.

[0071] For ease of understanding, some of the technical terms used in this application are explained below.

[0072] Cognitive diagnosis: Cognitive diagnosis refers to the diagnosis and evaluation of an individual's knowledge structure, processing skills, or cognitive processes (collectively referred to as attributes). Specifically, in intelligent education systems, the purpose of cognitive diagnosis is to diagnose a student's proficiency with specific knowledge concepts based on their response records. For example... Figure 1 As shown, Student A practiced a series of exercises, and Student A's response matrix is ​​as follows. Figure 1 As shown in (a), this indicates the student A's answer record for the exercises; the knowledge point association matrix between the exercises and the knowledge points involved in the exercises is as follows. Figure 1 As shown in (b) in the middle; then Figure 1 (a) and Figure 1 The information provided in (b) is used to obtain student A's proficiency level regarding the corresponding knowledge point through cognitive diagnosis. In one exemplary scheme, student A's proficiency level regarding the corresponding knowledge point can be determined through methods such as... Figure 1 It is displayed in the manner shown in (c) above.

[0073] Adaptive question selection: Adaptive question selection refers to recommending exercises to students from a question bank that meet certain specific objectives based on their cognitive diagnostic results. These specific objectives could be improving students' proficiency in weak knowledge points, or improving their mastery of new knowledge points, etc. The specific process can be as follows: Figure 2As shown, the student's answer record is "Question e1 answered correctly, Question e2 answered incorrectly, Question e3 answered incorrectly, Question e4 answered correctly, Question e5 not answered"; then, the above answer record and the knowledge point correlation matrix of Questions e1 to e5 are used to obtain the student's diagnostic result through cognitive diagnosis, where the diagnostic result can be expressed as "Poor mastery of knowledge points 1, 2 and 3; Questions corresponding to knowledge points 4 and 5 have not been done; Knowledge points 6 and 7 have been mastered well"; then, based on the diagnostic result, "Questions e5, e6, e7 and e8" are adaptively generated from the question bank.

[0074] Backpropagation: Forward propagation refers to the feedforward processing of the model, while backpropagation is the opposite of forward propagation, referring to updating the weight parameters of each layer of the model based on the model's output. For example, if the model includes an input layer, hidden layers, and an output layer, then forward propagation means processing the data in the order of input layer-hidden layer-output layer, while backpropagation means updating the weight parameters of each layer sequentially in the order of output layer-hidden layer-input layer.

[0075] This application provides a method, apparatus, device, and storage medium for recommending exercises, which can flexibly recommend exercises to a target object, thereby achieving wide applicability of the exercise recommendation method and meeting various application scenarios. The following describes exemplary applications of the electronic device provided in this application. The electronic device provided in this application can be implemented as various types of user terminals or as a server.

[0076] By running the exercise recommendation scheme provided in the embodiments of this application, the electronic device can flexibly recommend exercises to the target object, thereby realizing the wide applicability of the exercise recommendation method and meeting many application scenarios. That is, it improves the exercise recommendation performance of the electronic device itself and is applicable to multiple application scenarios of exercise recommendation.

[0077] See Figure 3 , Figure 3This is an optional architecture diagram of an application scenario for the exercise recommendation scheme provided in this application embodiment. To support an exercise recommendation application, terminal device 100 connects to server 300 via network 200, and server 300 connects to database 400. Network 200 can be a wide area network (WAN), a local area network (LAN), or a combination of both. The client for implementing the exercise recommendation scheme is deployed on terminal device 100. The client can run on terminal device 100 via a browser or as a standalone application (APP). The specific form of the client is not limited here. Server 300 involved in this application can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Terminal device 100 can be a smartphone, tablet, laptop, PDA, personal computer, smart TV, smartwatch, in-vehicle device, wearable device, etc., but is not limited to these. Terminal device 100 and server 300 can be directly or indirectly connected via network 200 through wired or wireless communication, which is not limited in this application. The number of server 300 and terminal device 100 is also not limited. The solution provided in this application can be completed independently by terminal device 100, independently by server 300, or jointly by terminal device 100 and server 300, which is not specifically limited in this application. In short, database 400 can be regarded as an electronic filing cabinet—a place to store electronic files, where users can perform operations such as adding, querying, updating, and deleting data in the files. The so-called "database" is a collection of data stored together in a certain way, which can be shared by multiple users, has the lowest possible redundancy, and is independent of applications. Database management system (DBMS) is a computer software system designed for managing databases, and generally has basic functions such as storage, retrieval, security, and backup.Database management systems can be categorized based on the database model they support, such as relational or Extensible Markup Language (XML); or based on the type of computer they support, such as server clusters or mobile phones; or based on the query language used, such as Structured Query Language (SQL) or XQuery; or based on performance priorities, such as maximum scale or highest operating speed; or other classification methods. Regardless of the classification method used, some DBMSs can cross categories, for example, supporting multiple query languages ​​simultaneously. In this application, database 400 can be used to store a question bank or the target object's historical answer records for X questions, as well as the knowledge point association information of the questions. Of course, the storage location of the question bank or the target object's historical answer records for X questions, as well as the knowledge point association information of the questions, is not limited to a database; for example, it can also be stored in terminal device 100, a blockchain, or the distributed file system of server 300, etc.

[0078] In some embodiments, server 300 can execute the exercise recommendation method and the cognitive diagnostic model training method provided in this application embodiment. In this embodiment, the server executes the cognitive diagnostic model training method before executing the exercise recommendation method. The specific process is as follows: obtain the historical answer records of X exercises for the target object from the terminal device 100 and / or database 400, as well as the knowledge point information involved in each of the X exercises; then construct the feature representation of the target object based on the historical answer records and knowledge point information; input the feature representation into the initial cognitive diagnostic model to obtain the predicted response of the target object to the X exercises; use cross-entropy to determine the loss value between the predicted response and the actual response of the target object to the X exercises; and adjust the initial weight parameters of the initial cognitive diagnostic model in reverse according to the loss value, thereby training the cognitive diagnostic model. The weight parameters of the cognitive diagnostic model include the knowledge relevance between the knowledge points involved in the exercises, the target object's knowledge proficiency in the knowledge points involved in the exercises, the difficulty of the exercises for the target object's knowledge points, and the knowledge point discrimination between the knowledge points involved in the exercises.

[0079] After the cognitive diagnostic model is trained, the server 300 can save it locally to provide remote exercise recommendation functionality to the terminal device 100. For example, the server 300 inputs the feature representation of the target object into the cognitive diagnostic model to obtain the predicted scores of each exercise in the exercise library, and then determines the candidate exercises corresponding to the target object based on the predicted scores; finally, it constructs a multi-objective optimization problem and its constraints based on the weight parameters in the cognitive diagnostic model, and solves the multi-objective optimization problem based on the constraints to determine the recommended exercises for the target object from the candidate exercises; finally, it sends the recommended exercises to the terminal device 100 so that the terminal device 100 can display the recommended exercises in the graphical interface 110.

[0080] Server 300 can also train the completed cognitive diagnostic model and send (deploy) it to terminal device 100, thereby enabling object detection locally on terminal device 100. For example, terminal device 100 inputs the feature representation of the target object into the cognitive diagnostic model to obtain the predicted scores of each exercise in the exercise library, and then determines the candidate exercises corresponding to the target object based on the predicted scores; finally, it constructs a multi-objective optimization problem and its constraints based on the weight parameters in the cognitive diagnostic model, and solves the multi-objective optimization problem based on the constraints to determine the recommended exercises for the target object from the candidate exercises; finally, it displays the recommended exercises on the graphical interface 110 of terminal device 100.

[0081] It is understood that in the specific implementation of this application, data such as historical answer records and knowledge point information are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0082] Based on the above introduction, the following section describes the exercise recommendation method in this application, taking the terminal device as the executing entity. Please refer to [link / reference]. Figure 4 One embodiment of the exercise recommendation method in this application includes:

[0083] 401. Obtain the historical answer records of X exercises for the target object, as well as the knowledge point information involved in each of the X exercises, where X is a positive integer.

[0084] In this embodiment, the terminal device is deployed with a client for recommending practice questions. When a user runs the client through the terminal device, the terminal device can obtain the target user's historical answer records for X practice questions, as well as the knowledge point information involved in each of the X practice questions. In one exemplary solution, the terminal device can obtain the target user's historical answer records from online practice question assessments or from offline exams or homework. Specifically, the historical answer records can be as follows: Figure 1 Recording and saving are performed as shown in (a). The knowledge points involved in each of the X exercises can be compared with, for example... Figure 1 Record and save as shown in (b).

[0085] In one exemplary scenario, assume the target is student A, and the question bank contains a total of 10 questions. Student A's current historical answer records are shown in Table 1:

[0086] Table 1

[0087]

[0088] The knowledge point matrix corresponding to the 10 exercises in this question bank is shown in Table 2:

[0089] Table 2

[0090]

[0091] In Table 2, a value of 1 between a knowledge point and an exercise indicates that the knowledge point is involved in the exercise; a value of 0 between a knowledge point and an exercise indicates that the knowledge point is not involved in the exercise.

[0092] 402. Construct a feature representation of the target object based on the historical answer record and the knowledge point information.

[0093] In this embodiment, after obtaining the historical answer records of the target object, the terminal device combines these historical answer records with the knowledge point information to construct a feature representation of the target object. As shown in Tables 1 and 2 in step 401, after obtaining student A's answer records (Table 1), combining them with the knowledge point matrix shown in Table 2 will construct a feature representation of student A. It is understood that this feature representation can be a matrix or other forms; no specific limitation is made here, as long as it can represent the characteristics of student A.

[0094] 403. Input the feature representation into the cognitive diagnostic model so that the cognitive diagnostic model outputs candidate exercises.

[0095] In this embodiment, the terminal device needs to train an initial cognitive diagnostic model based on the feature representation to obtain the cognitive diagnostic model. The weight parameters of the initial cognitive diagnostic model include initial knowledge proficiency, initial knowledge relevance, initial knowledge point difficulty, and initial knowledge point discrimination. The feature representation is input into the cognitive diagnostic model to obtain the predicted score of each exercise in the exercise bank. The candidate exercise is output from the exercise bank based on the predicted score.

[0096] Specifically, the process by which the terminal device obtains the cognitive diagnosis based on the feature and by training an initial cognitive diagnostic model can be as follows: Establishing the initial cognitive diagnostic model; passing the feature representation through the first layer of the initial cognitive diagnostic model to obtain the hidden ability representation vector of the target object; passing the hidden ability representation vector through the second layer of the initial cognitive diagnostic model to obtain the initial knowledge proficiency; passing the feature representation through the exercise knowledge point association matrix in the initial cognitive diagnostic model to obtain the initial knowledge relevance; and passing the feature representation through the exercise difficulty training matrix in the initial cognitive diagnostic model to obtain the initial knowledge difficulty. The feature representation is used to obtain the initial knowledge point discrimination through the exercise discrimination training matrix in the initial cognitive diagnostic model; the predicted response corresponding to the X exercises is calculated based on the initial knowledge proficiency, the initial knowledge relevance, the initial knowledge point difficulty, and the initial knowledge point discrimination; the loss value of the initial cognitive diagnostic model is obtained by using cross-entropy based on the predicted response and the actual response of the X exercises; the weight parameters of the initial cognitive diagnostic model are adjusted in reverse based on the loss value to obtain the cognitive diagnostic model, and the weight parameters of the cognitive diagnostic model are the knowledge proficiency, the knowledge relevance, the knowledge point difficulty, and the knowledge point discrimination.

[0097] In one exemplary scheme, the cognitive diagnostic model can be as follows: Figure 5 As shown, this cognitive diagnostic model includes two networks (a hidden capability representation network layer and a fully connected network layer) to model the knowledge proficiency of the target object. The model includes a knowledge point association matrix, a question difficulty training matrix, and a question discrimination training matrix. These training matrices are then used to diagnose the question one-hot vectors in the feature representation, obtaining knowledge relevance, knowledge point difficulty, and knowledge point discrimination. Knowledge proficiency, knowledge relevance, knowledge point difficulty, and knowledge point discrimination are used as weight parameters of the cognitive diagnostic model. The input to the cognitive diagnostic model is the feature representation (i.e., the student's one-hot vector and the question's one-hot vector). Then, based on the aforementioned weight parameters, the model predicts the student's estimated response to the questions already answered. The model then calculates a loss value based on the estimated response and the student's actual response to the questions already answered. Finally, the model is trained by adjusting the weight parameters in reverse based on the loss value.

[0098] In this cognitive diagnostic model, the specific operation of modeling the knowledge proficiency of the target object through two networks (i.e., the hidden capability representation network layer and the fully connected network layer) can be as follows: In the first layer network (e.g., Figure 5 In the hidden capability representation network layer shown, we use formula (1) to represent the student's one-hot representation vector. Mapped to its hidden capability representation vector e i , where W1∈R d*L d is e i The size. In the second layer network (such as...) Figure 5 In the fully connected network layer shown in Equation (2), we use a positive fully connected layer to hide the capability representation vector e. i Mapped to students' knowledge proficiency α i Formulas (1) and (2) are as follows:

[0099]

[0100] α i =Φ(W2×(e i ) T +b2) (2)

[0101] Where W1 and W2 are the weight matrices of the network, and b1 and b2 are constants.

[0102] The loss function of this cognitive diagnostic model during training is defined as the predicted response y. ij With the actual response r ij The cross-entropy between them is shown in Equation (3).

[0103]

[0104] During training, after calculating the loss value, the knowledge proficiency, knowledge relevance, knowledge difficulty, and knowledge discrimination are adjusted in reverse based on the loss value.

[0105] 404. Obtain the target parameters, which are the weight parameters of the cognitive diagnostic model. The weight parameters include the knowledge relevance between the knowledge points involved in the exercises, the target object's knowledge proficiency in the knowledge points involved in the exercises, the difficulty of the exercises for the target object's knowledge points, and the knowledge point differentiation between the knowledge points involved in the exercises.

[0106] After the terminal device acquires the trained cognitive diagnostic model, it uses the weight parameters of the cognitive diagnostic model as the target parameters. In the trained cognitive diagnostic model, the weight parameters will include the knowledge relevance between the knowledge points involved in the exercises, the target object's proficiency in the knowledge points involved in the exercises, the difficulty of the exercises for the target object's knowledge points, and the knowledge point differentiation between the knowledge points involved in the exercises.

[0107] 405. Construct a multi-objective optimization problem and its constraints based on the objective parameters. The constraints include at least the objective object's proficiency with the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the difficulty of the knowledge points in the recommended exercises reaching a first threshold and the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaching a second threshold.

[0108] After obtaining the target parameters, the terminal device can construct a multi-objective optimization problem and its constraints based on preset requirements. In this embodiment, the constraints at least include the target object's proficiency with the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem at least include the difficulty of the knowledge points in the recommended exercises reaching a first threshold and the knowledge relevance between the knowledge points in the recommended exercises and the knowledge points in the X exercises reaching a second threshold. It can be understood that the constraints may also include the number of recommended exercises and the constraint that the recommended exercises are unanswered by the target object. The objectives may also include the number of knowledge points involved in the recommended exercises reaching a third threshold and the number of recommended exercises reaching a fourth threshold.

[0109] Based on the above requirements, the terminal device can construct the multi-objective optimization problem and the constraints in the following manner: It constructs the first objective and first constraints in the multi-objective optimization problem based on the knowledge proficiency in the objective parameters. The first objective is that the difficulty of the knowledge points in the recommended exercises reaches a first threshold, and the first constraint is the knowledge proficiency constraint of the target object regarding the knowledge points involved in the recommended exercises. It constructs the second objective and second constraints in the multi-objective optimization problem based on the knowledge relevance in the objective parameters. The second objective is that the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaches a second threshold, and the second constraint is the number of knowledge points involved in the recommended exercises.

[0110] It is understandable that the second objective in this embodiment is to improve the target audience's proficiency with knowledge points. During the recommendation process, recommendations for knowledge points that the target audience has already mastered well can be reduced, while recommendations for knowledge points that the target audience has mastered poorly can be increased. That is, it is necessary to ensure that the correlation between recommended exercises and knowledge points that the target audience has mastered well is low, while the correlation between recommended exercises and knowledge points that the target audience has mastered poorly is high. When constructing this knowledge point relevance objective, this embodiment can use the Pearson correlation coefficient to measure the above knowledge relevance, where the Pearson correlation coefficient can be expressed as formula (4):

[0111]

[0112] Where α i It is a real vector of length K, used to represent the knowledge proficiency of the target object; and σα i α i The mean and standard deviation of n; i This is used to indicate the number of exercises associated with each knowledge point. O1(n) i α i The smaller the value, the higher the relevance, which further indicates that the recommended exercises cover more knowledge points that the target audience has a poor grasp of. Therefore, we will use O1(n) i α i As the second objective.

[0113] In this embodiment, the goal is that when the number of knowledge points involved in the recommended exercises reaches a third threshold, the recommended exercises should include knowledge points that the target student has not answered. These knowledge points should be related to the knowledge points the student has already mastered. The number of knowledge points can be represented as shown in Formula 5:

[0114]

[0115] in It is element-wise multiplication, sum is the sum operation, (n i ≠0)∈{0,1} 1*K It is based on logical operations and outputs a binary vector, representing whether the recommended exercises contain the corresponding knowledge points. i ∈{0,1} 1*K This indicates whether the target user has answered the corresponding knowledge point. O2(n i The larger the value of ), the more new knowledge points are included in the recommended exercises, which can encourage the target audience to explore new knowledge.

[0116] To further enhance the diversity of knowledge points in the recommended exercises, a goal of diversity is proposed, which can be defined as shown in formula (6):

[0117]

[0118] O3(n i The larger the value, the more knowledge points are included in the recommended exercises.

[0119] Based on the above objectives, the constructed multi-objective optimization problem can be represented as shown in Equation 7:

[0120]

[0121] Furthermore, due to the characteristics of the exercise recommendation problem, the above multi-objective optimization problem also has multiple constraints, and the final form can be shown in formula (8):

[0122]

[0123] Among them, v i ∈{0,1} 1*M Used to indicate whether the target object has completed the corresponding exercises. This is used to indicate the probability that the target object will answer the practice questions correctly. The first constraint C1 indicates that the number of practice questions selected during the recommendation process is equal to m. The second constraint C2 indicates that practice questions that the target object has already completed will not be selected. The third constraint C3 indicates that the difficulty of the knowledge points of each practice question selected during the recommendation process reaches a preset threshold. The final constraint C4 is used to ensure that the overall accuracy rate of students in the recommended practice group is high.

[0124] 406. Solve the multi-objective optimization problem according to the at least two constraints, so as to obtain the recommended problem from the candidate problems.

[0125] After the terminal device constructs the multi-objective optimization problem and its constraints based on the target parameters, it solves the multi-objective optimization problem according to at least two constraints and the candidate exercises to obtain the optimal individual, and then uses this optimal individual as the recommended exercise. It is understood that the recommended exercise is a combination of exercises, meaning it includes multiple exercises.

[0126] In this embodiment, the terminal device can use a dual-encoding and dual-population co-evolutionary algorithm to solve the multi-objective optimization problem from the candidate problems, so that the problems that satisfy the at least two constraints are used as the recommended problems.

[0127] Optionally, when solving the multi-objective optimization problem, the terminal device can construct the initial population of the two populations based on the candidate exercises. To accelerate convergence, when constructing the main population and auxiliary population based on the candidate exercises, the terminal device ensures that the main population includes initial population individuals oriented towards diversity evolution, and the auxiliary population includes initial population individuals oriented towards convergence evolution. In one implementation, the terminal device uses binary encoding to generate exercise combinations to obtain the initial population individuals of the main population for the candidate exercises, and uses integer encoding of the knowledge points involved in the candidate exercises to generate exercise combinations to obtain the initial population individuals of the auxiliary population. An exemplary scheme shows the encoding diagram of the terminal device constructing the main population and auxiliary population based on the candidate exercises as follows: Figure 6 As shown. Assume that the information about each exercise in the question bank and the knowledge points involved in that exercise can be represented as follows: Figure 6 As shown in (a), the recommended number of exercises is 5, and the integer codes of the knowledge points involved are (0, 3, 3, 2, 2). That is, the recommended exercises need to involve 0 knowledge points 1, 3 knowledge points 2, 3 knowledge points 3, 2 knowledge points 4, and 2 knowledge points 5. At this time, the individuals in this auxiliary population can be as follows: Figure 6 As shown in (b), the integer encoding of the knowledge point is (0, 3, 3, 2, 2); and individuals in this main population can be as follows: Figure 6 As shown in (c), the binary encoding is (1, 1, 0, 0, 0, 1, 0, 1, 0, 1).

[0128] During the solution process, the integer encoding of knowledge points can be decoded into multiple lists of recommended exercises. This makes the group of knowledge point integer encodings less likely to get stuck in local optima and allows them to explore new regions. Therefore, the knowledge point integer encoding can be used to assist the group in performing a coarse search, responsible for finding the optimal distribution of knowledge point exercises, thus providing fast convergence. In an exemplary scheme, where the integer encoding of a knowledge point is (0, 3, 3, 2, 2), the decoded recommended exercises can be as follows:

[0129] Recommended Exercise List 1: {Exercise 1, Exercise 2, Exercise 6, Exercise 8, Exercise 10};

[0130] Recommended Exercise List 2: {Exercise 1, Exercise 2, Exercise 6, Exercise 7, Exercise 8};

[0131] Recommended Exercise List 3: {Exercise 2, Exercise 5, Exercise 6, Exercise 8, Exercise 10}.

[0132] The main population uses binary encoding to generate individuals, with each binary code corresponding to a recommended list of exercises. This allows the main population to perform a fine-grained search, determining which exercises are selected as recommended, thus providing good diversity of exercises. In an exemplary scheme, if an individual in the main population is encoded as (1, 1, 0, 0, 0, 1, 0, 1, 0, 1), then its corresponding recommended list of exercises is {Exercise 1, Exercise 2, Exercise 6, Exercise 8, Exercise 10}.

[0133] After the terminal device constructs a main population and an auxiliary population according to the above method, it performs co-evolution on these two main populations and the auxiliary population, that is, it uses the auxiliary population to guide the evolution of the main population. Specifically, it can be as follows: A main population and an auxiliary population are constructed based on the candidate exercises. The main population includes initial population individuals oriented towards diversity evolution, and the auxiliary population includes initial population individuals oriented towards convergence evolution. An auxiliary profile is obtained by performing optimization calculations on the parent individuals in the auxiliary population, and the auxiliary profile includes optimized individuals obtained from the optimization calculations on the auxiliary population. The optimized individuals corresponding to the main population are obtained by performing optimization calculations on the main population based on the guiding individuals in the auxiliary profile and the parent individuals of the main population. The recommended exercises are obtained based on the optimized individuals of the auxiliary population and the optimized individuals of the main population. That is, the specific process can be as follows: Figure 7 As shown: First, two populations are initialized using different initialization strategies. Then, a binary competition selection process is used to mate individuals from the main population and the auxiliary population respectively. Next, the auxiliary population performs offspring generation, individual evaluation, environmental selection, and archive updates, while the main population waits for the auxiliary population to update its auxiliary archive, which retains the best individuals from that auxiliary population in each optimization process. Next, the mating parents in the main population utilize individuals from the auxiliary archive through a guidance strategy. Finally, environmental selection is applied to both populations to preserve potentially good individuals. Steps two through six are repeated until the evolutionary termination criterion is met, after which the best individual is output as a recommended exercise.

[0134] In this embodiment, the specific process of the auxiliary population evolving the main population according to the guidance strategy can be as follows: Select any one individual from the auxiliary file as the guidance individual and obtain the knowledge point integer code of the guidance individual; select any two individuals from the main population as the parent individuals and obtain the knowledge point integer codes of the parent individuals; calculate the difference in the number of exercises corresponding to the knowledge points between the guidance individual and the parent individual by comparing the knowledge point integer codes of the guidance individual and the parent individual; calculate the flipping probability of each knowledge point in the parent individual based on the difference in number, and flip each knowledge point in the parent individual according to the flipping probability to obtain a crossover individual; determine the flipping flag of each knowledge point in the crossover individual based on the flipping probability and a first random value, and flip each knowledge point in the crossover individual according to the flipping flag to obtain a mutated individual, and use the mutated individual as the offspring individual; execute the above process sequentially until the optimization calculation is completed and the optimized individual corresponding to the main population is output.

[0135] When the terminal device evolves the main population according to the guidance strategy of the auxiliary population, in one exemplary scheme, it is assumed that a parent individual in the main population is I. off One of the guiding individuals in this auxiliary file is I. guidance Their knowledge points are represented by integer codes as IntEn(I off ) and IntEn(I guidance Then, I off and I guidance The difference in the number of practice questions for each knowledge point can be calculated using formula (9):

[0136]

[0137] Where Diff1 represents I off and I guidance The difference in the number of exercises corresponding to knowledge point 1 between them, and so on.

[0138] Assume there are N questions in the question bank kj Each exercise is related to knowledge point j, I off Choose A set of exercises related to knowledge point j, I guidance have We have a set of exercises on knowledge point j. Then, we propose the following two probabilities of flipping, which can be shown in formula (10):

[0139]

[0140] Where P c This is the crossover probability, which is usually set to 0.5. It is to I off The probability that the flag bit related to knowledge point j flips from zero to one. This refers to I off The probability that the flag related to knowledge point j will flip from one to zero. Next, two random vectors are generated. and To determine which 0 or 1 bits related to knowledge point j need to be flipped, the specific calculation process can be shown in formula (11):

[0141]

[0142] in and There are two flip flags, and a flag value of 1 means to perform a flip.

[0143] Based on the above description, the operational process of this terminal device in co-evolution according to the guidance strategy can be as follows: Figure 8 As shown: First, two individuals are selected from the parent population in the mating pool of the main population. Then, a random number (i.e., the second random value) is generated and compared with the proportion parameter. If the random number is greater than the proportion parameter, single-point crossover and mutation are applied to the parent individuals; otherwise, a guide individual is randomly selected from the auxiliary files of the auxiliary population. Next, the difference between the parent individuals and the guide individuals is calculated according to the above formula (9). Then, the flip probability and flip flag of each decision variable bit of the parent individuals are calculated using the above formulas (10) and (11), and the corresponding operations are finally performed. Finally, the guided parent individuals are added to the offspring population. Finally, after optimization by the above-proposed guide-based dual-population dual-encoding co-evolutionary algorithm, the final individual is output, and the optimal individual is selected as the final solution, i.e., the recommended exercise.

[0144] In this embodiment, experiments are conducted on two widely used datasets, ASSISTments and JunYi, to verify the effectiveness of our proposed method. Table 3 describes these two datasets:

[0145] Table 3

[0146]

[0147]

[0148] In this experiment, the goal was to help students consolidate their weaknesses. The better a student's proficiency with a particular knowledge point, the fewer related practice questions could be recommended. In other words, the lower the correlation between a student's proficiency with a knowledge point and the knowledge points covered in the recommended practice questions, the better the quality of the recommended questions. To measure this correlation, three widely used metrics were employed: Pearson correlation coefficient, Spearman correlation coefficient, and Kendall correlation coefficient. The final comparison results are shown in Table 4 below.

[0149] Table 4

[0150]

[0151] As can be seen from Table 4, the technical solution provided in this application achieves the best results on different datasets and different metrics. Therefore, the technical solution provided in this application can effectively recommend suitable exercise combinations for students, thereby helping them consolidate their weak knowledge points.

[0152] In addition to the goal of reinforcing weaknesses, we also compared the diversity and novelty of the recommended exercises. The final comparison results are shown in Table 5 below:

[0153] Table 5

[0154]

[0155]

[0156] As can be seen from Table 5, the technical solution provided in this application is superior to other methods in terms of the diversity and novelty of the recommended exercise combinations. This demonstrates that the technical solution provided in this application achieves good results across multiple objectives, possesses high flexibility, and can meet the needs of various practical scenarios.

[0157] Based on the above technical description, the specific application scenarios of the technical solution provided in this application are described below, which are mainly divided into two categories:

[0158] I. Student-oriented products such as tablets are designed for individual students, providing practice exercises during the synchronous learning phase and exam preparation during the review phase. Assuming the terminal device is a smart tablet and the target audience is students, the interface display during student review and exam preparation could be as follows: Figure 9a He Ru Figure 9b As shown. Among them, Figure 9a Recommendations for practice problems for this student during their learning phase. Figure 9b Recommend practice questions for the student's review and exam preparation.

[0159] Second, products for schools and teachers, including personalized assignments given by teachers to suit the individual circumstances of each student.

[0160] Assuming the terminal device is a smart tablet, the target audience is students, and the teacher recommends test papers to students, the interface could be displayed as follows: Figure 10 As shown. Among them, Figure 10 This is a recommendation for homework assignments given by the teacher to students during their periodic learning.

[0161] Understandably, the interface for selecting recommended exercises between the student and the teacher is as follows: Figure 11 As shown, this includes the selection of question type, number of questions, and difficulty level.

[0162] The exercise recommendation device in this application is described in detail below. Please refer to [link / reference]. Figure 12 , Figure 12 This is a schematic diagram of one embodiment of the exercise recommendation device in this application. The exercise recommendation device 20 includes:

[0163] The acquisition module 201 is used to acquire the historical answer records of X exercises of the target object, as well as the knowledge point information involved in each of the X exercises, where X is a positive integer;

[0164] The processing module 202 is used to construct a feature representation of the target object based on the historical answer record and the knowledge point information; and input the feature representation into the cognitive diagnostic model so that the cognitive diagnostic model outputs candidate exercises.

[0165] The acquisition module 201 is also used to acquire target parameters, which are weight parameters of the cognitive diagnostic model. The weight parameters include the knowledge relevance between the knowledge points involved in the exercises, the knowledge proficiency of the target object in the knowledge points involved in the exercises, the difficulty of the exercises for the target object in the knowledge points, and the knowledge point differentiation between the knowledge points involved in the exercises.

[0166] The processing module 202 is further configured to construct a multi-objective optimization problem and constraints on the multi-objective optimization problem based on the target parameters. The constraints include at least the target object's knowledge proficiency in the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the difficulty of the knowledge points in the recommended exercises reaching a first threshold and the knowledge correlation between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaching a second threshold.

[0167] Output module 203 is used to solve the multi-objective optimization problem based on the at least two constraints, so as to obtain the recommended problem from the candidate problems.

[0168] This application provides a problem recommendation device. Using this device, the target object's knowledge mastery is obtained through a cognitive diagnostic model. Then, based on the target object's knowledge mastery, a corresponding multi-objective optimization problem and its constraints are constructed. Finally, the multi-objective optimization problem is solved to obtain recommended problems. Since the multi-objective optimization problem can be flexibly adjusted according to actual conditions, problems can be flexibly recommended to the target object, thus achieving wide applicability of the problem recommendation method and meeting various application scenarios.

[0169] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application,

[0170] The processing module 202 is specifically used to train an initial cognitive diagnostic model based on the feature representation to obtain the cognitive diagnostic model. The weight parameters of the initial cognitive diagnostic model include initial knowledge proficiency, initial knowledge relevance, initial knowledge point difficulty, and initial knowledge point discrimination.

[0171] Inputting this feature representation into the cognitive diagnostic model yields the predicted score for each exercise in the exercise bank;

[0172] The candidate exercise is output from the exercise bank based on the predicted score.

[0173] This application provides a problem recommendation device. Using this device, for each target object, a cognitive diagnostic model corresponding to the target object is obtained based on its feature representation, and candidate problems corresponding to the target object are obtained based on the cognitive diagnostic model. In this way, based on the target object's profile information, targeted problem recommendations can be made, improving the accuracy of the problem recommendations.

[0174] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the processing module 202 is specifically used to establish the initial cognitive diagnosis model;

[0175] The hidden capability representation vector of the target object is obtained by passing this feature representation through the first layer network of the initial cognitive diagnostic model;

[0176] The hidden ability representation vector is passed through the second layer of the initial cognitive diagnostic model to obtain the initial knowledge proficiency.

[0177] The feature representation is used to obtain the initial knowledge relevance through the exercise knowledge point association matrix in the initial cognitive diagnostic model, the feature representation is used to obtain the initial knowledge point difficulty through the exercise difficulty training matrix in the initial cognitive diagnostic model, and the feature representation is used to obtain the initial knowledge point discrimination through the exercise discrimination training matrix in the initial cognitive diagnostic model.

[0178] Calculate the predicted response for each of the X exercises based on the initial knowledge proficiency, the initial knowledge relevance, the initial knowledge difficulty, and the initial knowledge discrimination.

[0179] The loss value of the initial cognitive diagnostic model is obtained using cross-entropy based on the predicted response and the actual responses of the X exercises.

[0180] The initial cognitive diagnostic model is obtained by adjusting the weight parameters of the cognitive diagnostic model in reverse based on the loss value. The weight parameters of the cognitive diagnostic model are the knowledge proficiency, the knowledge relevance, the knowledge difficulty, and the knowledge discrimination.

[0181] In this embodiment, a problem recommendation device is provided. Using this device, during training, the knowledge proficiency, knowledge relevance, knowledge differentiation, and knowledge difficulty between the target object and the problems are used as weight parameters of the cognitive diagnostic model. Simultaneously, a two-layer network representation is applied to the feature representation of the target object, making the target object's answer records more strongly correlated with the information of the knowledge points, thereby improving the recommendation accuracy of the cognitive diagnostic model and ultimately improving the accuracy of the problem recommendation.

[0182] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the output module 203 is specifically used to solve the multi-objective optimization problem from the candidate exercises using a dual-encoding and dual-population co-evolutionary algorithm, so that the exercises that satisfy the at least two constraints are used as the recommended exercises.

[0183] This application provides a problem recommendation device. Using this device, a dual-encoding and dual-population co-evolutionary algorithm is employed to solve the multi-objective optimization problem, thereby achieving rapid convergence and accelerating the solution process.

[0184] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the output module 203 is specifically used to construct a main population and an auxiliary population according to the candidate exercises. The main population includes initial population individuals oriented towards the direction of diversity evolution, and the auxiliary population includes initial population individuals oriented towards the direction of convergence evolution.

[0185] An auxiliary profile is obtained by performing optimization calculations on the parent individuals in the auxiliary population. The auxiliary profile includes the optimized individuals obtained from the optimization calculations on the auxiliary population.

[0186] Based on the guiding individuals in the auxiliary file and the parent individuals of the main population, the optimization calculation of the main population is performed to obtain the optimized individuals corresponding to the main population;

[0187] The recommended exercise is derived from the optimal individuals of the auxiliary population and the optimal individuals of the main population.

[0188] In this embodiment, a problem recommendation device is provided. Using this device, the main population includes initial population individuals oriented towards diversity evolution, and the auxiliary population includes initial population individuals oriented towards convergence evolution. This allows the main population to perform evolutionary calculations based on the auxiliary population, thereby achieving rapid convergence and accelerating the solution process.

[0189] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the processing module 202 is specifically used to generate exercise combinations from the candidate exercises using binary encoding to obtain the initial population individuals of the main population, and to generate exercise combinations from the knowledge points involved in the candidate exercises using knowledge point integer encoding to obtain the initial population individuals of the auxiliary population.

[0190] In this embodiment, a problem recommendation device is provided. Using this device, individuals in the auxiliary population are constructed using knowledge concept integer encoding. Since the knowledge concept integer encoding can be decoded into multiple problem sets, the auxiliary population is less prone to getting trapped in local optima and can explore new regions. Therefore, the knowledge concept integer encoding is used by the auxiliary population to perform a coarse search, i.e., to search for the optimal distribution of knowledge concept exercises, thus providing fast convergence, while the binary encoding is used by the main population to perform a fine search, thus providing good diversity.

[0191] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the processing module 202 is specifically used to select any individual from the auxiliary file as the guidance individual and obtain the knowledge point integer code of the guidance individual;

[0192] Take any two individuals from the main population as the parent individuals, and obtain the knowledge point integer codes of the parent individuals;

[0193] The difference in the number of exercises corresponding to the knowledge points between the guided individual and the parent individual is calculated by comparing the integer codes of the knowledge points of the guided individual with those of the parent individual.

[0194] Calculate the flipping probability of each knowledge point in the parent individual based on the quantitative difference, and flip each knowledge point in the parent individual according to the flipping probability to obtain the crossover individual;

[0195] The flipping flag of each knowledge point in the crossover individual is determined based on the flipping probability and the first random value. The flipping flag is used to flip each knowledge point in the crossover individual to obtain a mutant individual, and the mutant individual is used as the offspring individual.

[0196] Repeat the above process until the optimization calculation is completed and the optimized individuals corresponding to the main population are output.

[0197] In this embodiment, a problem recommendation device is provided. Using this device, the main population includes initial population individuals oriented towards diversity evolution, and the auxiliary population includes initial population individuals oriented towards convergence evolution. This allows the main population to perform evolutionary calculations based on the auxiliary population, thereby achieving rapid convergence and accelerating the solution process.

[0198] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the processing module 202 is further used to generate a second random value;

[0199] The second random value is compared with the proportion parameter to obtain the comparison result. The proportion parameter is used to indicate the probability value of the main population receiving the guidance strategy of the auxiliary population.

[0200] When the comparison result indicates that the second random value is greater than or equal to the ratio parameter, the operation of performing optimization calculations on the main population based on the guide individuals in the auxiliary file and the parent individuals of the main population to obtain the optimized individuals corresponding to the main population is triggered.

[0201] When the comparison result indicates that the second random number is less than the ratio parameter, the optimization calculation of the main population is performed based on the parent individuals in the main population to obtain the optimized individuals.

[0202] This application provides a problem recommendation device. Using this device, multiple implementation methods are available, leading to more optimized evolutionary algorithms. The main population includes initial population individuals oriented towards diversity evolution, while the auxiliary population includes initial population individuals oriented towards convergence evolution. This allows the main population to perform evolutionary calculations based on the auxiliary population, achieving rapid convergence and accelerating the solution process.

[0203] Optionally, in the above Figure 12Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the processing module 202 is specifically used to arbitrarily select two parent individuals from the main population and flip the target knowledge points in the parent individuals to obtain crossover individuals;

[0204] Each knowledge point in the crossover individual is flipped according to the mutation probability to obtain a mutated individual, and this mutated individual is used as the offspring individual;

[0205] Repeat the above process until the optimization calculation is completed and the optimized individuals corresponding to the main population are output.

[0206] This application provides a problem recommendation device. Using this device, multiple evolutionary algorithm schemes are provided, increasing the feasibility of the schemes.

[0207] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the processing module 202 is specifically used to construct a first objective and a first constraint in the multi-objective optimization problem according to the knowledge proficiency in the target parameter. The first objective is that the knowledge point difficulty of the recommended exercise reaches a first threshold, and the first constraint is the knowledge proficiency constraint of the target object on the knowledge points involved in the recommended exercise.

[0208] Based on the knowledge relevance in the objective parameter, a second objective and a second constraint are constructed in the multi-objective optimization problem. The second objective is that the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaches a second threshold. The second constraint is a constraint on the number of knowledge points involved in the recommended exercises.

[0209] This application provides a problem recommendation device. Using this device, a corresponding multi-objective optimization problem and its constraints are constructed based on the target object's mastery of knowledge points. Finally, the multi-objective optimization problem is solved to obtain recommended problems. Since the multi-objective optimization problem can be flexibly adjusted according to actual conditions, problems can be flexibly recommended to the target object, thereby achieving wide applicability of the problem recommendation method and meeting various application scenarios.

[0210] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the constraints further include the constraint on the number of recommended exercises and the constraint that the recommended exercises are unanswered questions by the target object.

[0211] This application provides a problem recommendation device. By using this device, several other possible scenarios for the constraints are added, thereby increasing the feasibility of the solution.

[0212] Optionally, in the above Figure 12 Based on the corresponding embodiments, in another embodiment of the exercise recommendation device 20 provided in this application, the objectives of the multi-objective optimization problem also include the number of knowledge points involved in the recommended exercises reaching a third threshold and the number of recommended exercises reaching a fourth threshold.

[0213] This application provides a problem recommendation device. By using this device, several other possible scenarios for the target are added, increasing the feasibility of the solution.

[0214] The exercise recommendation device provided in this application can be used on a server; please refer to [link / reference]. Figure 13 , Figure 13 This is a schematic diagram of a server structure provided in an embodiment of this application. The server 300 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 322 (e.g., one or more processors) and memory 332, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 342 or data 344. The memory 332 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the server. Furthermore, the CPU 322 may be configured to communicate with the storage media 330 and execute the series of instruction operations stored in the storage media 330 on the server 300.

[0215] Server 300 may also include one or more power supplies 326, one or more wired or wireless network interfaces 350, one or more input / output interfaces 358, and / or one or more operating systems 341, such as Windows Server. TM Mac OS X TM Unix TM Linux TM FreeBSD TM etc.

[0216] The steps performed by the server in the above embodiments can be based on this Figure 13 The server structure shown.

[0217] The exercise recommendation device provided in this application can be used in terminal devices; please refer to [link / reference]. Figure 14For ease of explanation, only the parts relevant to the embodiments of this application are shown. For specific technical details not disclosed, please refer to the method section of the embodiments of this application. In the embodiments of this application, a smartphone is used as an example for illustration:

[0218] Figure 14 This is a block diagram illustrating a portion of the structure of a smartphone related to the terminal device provided in the embodiments of this application. (Reference) Figure 14 The smartphone includes components such as a radio frequency (RF) circuit 410, a memory 420, an input unit 430, a display unit 440, a sensor 450, an audio circuit 460, a wireless fidelity (WiFi) module 470, a processor 480, and a power supply 490. Those skilled in the art will understand that... Figure 14 The smartphone structure shown does not constitute a limitation on smartphones and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0219] The following is combined with Figure 14 A detailed introduction to the various components of a smartphone:

[0220] RF circuit 410 can be used for receiving and transmitting signals during information transmission or calls. Specifically, it receives downlink information from the base station and processes it with processor 480; additionally, it transmits uplink data to the base station. Typically, RF circuit 410 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier (LNA), a duplexer, etc. Furthermore, RF circuit 410 can also communicate wirelessly with networks and other devices. The aforementioned wireless communication can use any communication standard or protocol, including but not limited to Global System for Mobile Communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Message Service (SMS), etc.

[0221] The memory 420 can be used to store software programs and modules. The processor 480 executes various functions and data processing of the smartphone by running the software programs and modules stored in the memory 420. The memory 420 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the smartphone (such as audio data, phonebook, etc.). In addition, the memory 420 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0222] The input unit 430 can be used to receive input numerical or character information, and to generate key signal inputs related to user settings and function control of the smartphone. Specifically, the input unit 430 may include a touch panel 431 and other input devices 432. The touch panel 431, also known as a touch screen, can collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch panel 431), and drive the corresponding connected devices according to a pre-set program. Optionally, the touch panel 431 may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch position and the signal generated by the touch operation, and transmits the signal to the touch controller; the touch controller receives touch information from the touch detection device, converts it into touch point coordinates, and sends it to the processor 480, and can also receive and execute commands sent by the processor 480. In addition, the touch panel 431 can be implemented using various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel 431, the input unit 430 may also include other input devices 432. Specifically, other input devices 432 may include, but are not limited to, one or more of the following: physical keyboard, function keys (such as volume control buttons, power buttons, etc.), trackball, mouse, joystick, etc.

[0223] Display unit 440 can be used to display information input by the user or information provided to the user, as well as various menus of the smartphone. Display unit 440 may include display panel 441, optionally configured as a liquid crystal display (LCD), organic light-emitting diode (OLED), or similar form. Further, touch panel 431 may cover display panel 441. When touch panel 431 detects a touch operation on or near it, it transmits the information to processor 480 to determine the type of touch event. Subsequently, processor 480 provides corresponding visual output on display panel 441 based on the type of touch event. Although in Figure 14 In this embodiment, the touch panel 431 and the display panel 441 are two separate components to realize the input and output functions of the smartphone. However, in some embodiments, the touch panel 431 and the display panel 441 can be integrated to realize the input and output functions of the smartphone.

[0224] The smartphone may also include at least one sensor 450, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor can adjust the brightness of the display panel 441 according to the ambient light level, and the proximity sensor can turn off the display panel 441 and / or the backlight when the smartphone is moved to the ear. As a type of motion sensor, an accelerometer sensor can detect the magnitude of acceleration in various directions (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that recognize the smartphone's posture (such as landscape / portrait switching, related games, magnetometer posture calibration), vibration recognition-related functions (such as pedometer, tapping), etc. Other sensors that may be configured in the smartphone, such as gyroscopes, barometers, hygrometers, thermometers, and infrared sensors, will not be described in detail here.

[0225] Audio circuit 460, speaker 461, and microphone 462 provide an audio interface between the user and the smartphone. Audio circuit 460 converts received audio data into electrical signals and transmits them to speaker 461, where speaker 461 converts them into sound signals for output. On the other hand, microphone 462 converts collected sound signals into electrical signals, which are received by audio circuit 460, converted into audio data, and then processed by processor 480 before being transmitted via RF circuit 410 to, for example, another smartphone, or the audio data can be output to memory 420 for further processing.

[0226] WiFi is a short-range wireless transmission technology. Smartphones, through their WiFi modules (470), can help users send and receive emails, browse web pages, and access streaming media, providing wireless broadband internet access. Although Figure 14 WiFi module 470 is shown, but it is understood that it is not an essential component of a smartphone and can be omitted as needed without changing the nature of the invention.

[0227] The processor 480 is the control center of the smartphone, connecting various parts of the smartphone through various interfaces and lines. It performs various functions and processes data by running or executing software programs and / or modules stored in the memory 420, and by calling data stored in the memory 420, thereby providing overall monitoring of the smartphone. Optionally, the processor 480 may include one or more processing units; optionally, the processor 480 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the aforementioned modem processor may also not be integrated into the processor 480.

[0228] The smartphone also includes a power supply 490 (such as a battery) that supplies power to various components. Optionally, the power supply can be logically connected to the processor 480 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0229] Although not shown, smartphones may also include a camera, Bluetooth module, etc., which will not be described in detail here.

[0230] The steps performed by the terminal device in the above embodiments can be based on this Figure 14 The terminal device structure is shown.

[0231] This application also provides a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the methods described in the foregoing embodiments.

[0232] This application also provides a computer program product including a program, which, when run on a computer, causes the computer to perform the methods described in the foregoing embodiments.

[0233] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0234] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0235] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0236] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0237] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0238] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for recommending practice problems, characterized in that, include: Obtain the historical answer records of X exercises for the target object, as well as the knowledge point information involved in each of the X exercises, where X is a positive integer; Construct a feature representation of the target object based on the historical answer records and the knowledge point information; The feature representation is input into the cognitive diagnostic model so that the cognitive diagnostic model outputs candidate exercises; Obtain target parameters, which are weight parameters of the cognitive diagnostic model. The weight parameters include the knowledge relevance between the knowledge points involved in the exercises, the knowledge proficiency of the target object with the knowledge points involved in the exercises, the difficulty of the exercises for the target object's knowledge points, and the knowledge point differentiation between the knowledge points involved in the exercises. Based on the target parameters, a multi-objective optimization problem and constraints for the multi-objective optimization problem are constructed. The constraints include at least the target object's knowledge proficiency in the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the difficulty of the knowledge points in the recommended exercises reaching a first threshold and the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaching a second threshold. Solve the multi-objective optimization problem according to the at least two constraints, so as to obtain the recommended exercise from the candidate exercises; The step of inputting the feature representation into the cognitive diagnostic model so that the cognitive diagnostic model outputs candidate exercises includes: The initial cognitive diagnostic model is obtained by training an initial cognitive diagnostic model based on the feature representation. The weight parameters of the initial cognitive diagnostic model include initial knowledge proficiency, initial knowledge relevance, initial knowledge point difficulty, and initial knowledge point discrimination. The feature representation is input into the cognitive diagnostic model to obtain the predicted score for each exercise in the exercise bank; The candidate exercises are output from the exercise bank based on the predicted scores.

2. The method according to claim 1, characterized in that, The step of training the initial cognitive diagnostic model based on the feature representation to obtain the cognitive diagnostic model includes: Establish the initial cognitive diagnostic model; The feature representation is passed through the first layer network of the initial cognitive diagnostic model to obtain the hidden capability representation vector of the target object; The initial knowledge proficiency is obtained by passing the hidden ability representation vector through the second layer network of the initial cognitive diagnostic model. The feature representation is used to obtain the initial knowledge relevance through the exercise knowledge point association matrix in the initial cognitive diagnostic model, and the feature representation is used to obtain the initial knowledge point difficulty through the exercise difficulty training matrix in the initial cognitive diagnostic model, and the feature representation is used to obtain the initial knowledge point discrimination through the exercise discrimination training matrix in the initial cognitive diagnostic model; Calculate the predicted response for each of the X exercises based on the initial knowledge proficiency, the initial knowledge relevance, the initial knowledge difficulty, and the initial knowledge discrimination. The loss value of the initial cognitive diagnostic model is obtained using cross-entropy based on the predicted response and the actual responses of the X exercises. The cognitive diagnostic model is obtained by adjusting the weight parameters of the initial cognitive diagnostic model in reverse based on the loss value. The weight parameters of the cognitive diagnostic model are the knowledge proficiency, the knowledge relevance, the knowledge point difficulty, and the knowledge point discrimination.

3. The method according to claim 1, characterized in that, Solving the multi-objective optimization problem based on the at least two constraints, so as to obtain recommended exercises from the candidate exercises, includes: The multi-objective optimization problem is solved from the candidate problems using a dual-encoding and dual-population co-evolutionary algorithm, such that problems that satisfy at least two of the constraints are selected as recommended problems.

4. The method according to claim 3, characterized in that, The method of using a dual-encoding and dual-population co-evolutionary algorithm to solve the multi-objective optimization problem from the candidate problems, such that problems satisfying at least two constraints are selected as recommended problems, includes: Based on the candidate exercises, a main population and an auxiliary population are constructed. The main population includes initial population individuals oriented towards the direction of diversity evolution, and the auxiliary population includes initial population individuals oriented towards the direction of convergence evolution. An auxiliary profile is obtained by performing optimization calculations on the auxiliary population based on the parent individuals in the auxiliary population, and the auxiliary profile includes the optimized individuals obtained from the optimization calculations on the auxiliary population; Based on the guiding individuals in the auxiliary files and the parent individuals of the main population, the optimization calculation of the main population is performed to obtain the optimized individuals corresponding to the main population; The recommended exercises are obtained based on the optimized individuals of the auxiliary population and the optimized individuals of the main population.

5. The method according to claim 4, characterized in that, The construction of the main population and auxiliary population based on the candidate exercises includes: The candidate exercises are encoded using binary encoding to generate exercise combinations, which are then used to obtain the initial population individuals of the main population. The knowledge points involved in the candidate exercises are encoded using knowledge point integer encoding to generate exercise combinations, which are then used to obtain the initial population individuals of the auxiliary population.

6. The method according to claim 4, characterized in that, The step of performing optimization calculations on the main population based on the guiding individuals in the auxiliary archives and the parent individuals of the main population to obtain the optimized individuals corresponding to the main population includes: Select any individual from the auxiliary files as the guiding individual, and obtain the knowledge point integer code of the guiding individual; Take any two individuals from the main population as the parent individuals, and obtain the knowledge point integer codes of the parent individuals; The difference in the number of exercises corresponding to the knowledge points between the guiding individual and the parent individual is calculated by comparing the integer codes of the knowledge points of the guiding individual and the parent individual. Calculate the flipping probability of each knowledge point in the parent individual based on the quantitative difference, and flip each knowledge point in the parent individual according to the flipping probability to obtain a crossover individual; The flipping flag of each knowledge point in the cross individual is determined according to the flipping probability and the first random value, and the knowledge points in the cross individual are flipped according to the flipping flag to obtain the mutant individual, and the mutant individual is used as the offspring individual; The above process is executed sequentially until the optimization calculation is completed and the optimized individuals corresponding to the main population are output.

7. The method according to claim 4, characterized in that, Before performing optimization calculations on the main population based on the guide individuals in the auxiliary archives and the parent individuals of the main population to obtain the optimized individuals corresponding to the main population, the method further includes: Generate a second random value; The second random value is compared with the ratio parameter to obtain the comparison result. The ratio parameter is used to indicate the probability value of the main population receiving the guidance strategy of the auxiliary population. When the comparison result indicates that the second random value is greater than or equal to the ratio parameter, an operation is triggered to perform optimization calculations on the main population based on the guide individuals in the auxiliary file and the parent individuals of the main population to obtain the optimized individuals corresponding to the main population; When the comparison result indicates that the second random number is less than the ratio parameter, the optimization calculation of the main population is performed based on the parent individuals in the main population to obtain the optimized individuals.

8. The method according to claim 7, characterized in that, The step of performing optimization calculations on the main population based on parental individuals in the main population to obtain optimized individuals includes: Two parent individuals are randomly selected from the main population, and the target knowledge points in the parent individuals are flipped to obtain crossover individuals; Each knowledge point in the crossover individual is flipped according to the mutation probability to obtain a mutated individual, and the mutated individual is used as the offspring individual; The above process is executed sequentially until the optimization calculation is completed and the optimized individuals corresponding to the main population are output.

9. The method according to any one of claims 1 to 8, characterized in that, The multi-objective optimization problem is constructed based on the objective parameters, and at least two constraints are defined for the multi-objective optimization problem. Based on the knowledge point difficulty in the target parameters, a first objective and a first constraint are constructed in the multi-objective optimization problem. The first objective is that the knowledge point difficulty of the recommended exercises reaches a first threshold, and the first constraint is the knowledge proficiency of the target object with the knowledge points involved in the recommended exercises. Based on the knowledge relevance and knowledge proficiency in the target parameters, a second objective and a second constraint are constructed in the multi-objective optimization problem. The second objective is that the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaches a second threshold. The second constraint is a constraint on the number of knowledge points involved in the recommended exercises.

10. The method according to any one of claims 1 to 8, characterized in that, The constraints also include the number of recommended exercises and the constraint that the recommended exercises are those that the target object has not answered.

11. The method according to any one of claims 1 to 8, characterized in that, The objectives of the multi-objective optimization problem also include the number of knowledge points involved in the recommended exercises reaching a third threshold and the number of recommended exercises reaching a fourth threshold.

12. A problem recommendation device, characterized in that, include: The acquisition module is used to acquire the historical answer records of X exercises for the target object, as well as the knowledge point information involved in each of the X exercises; The processing module is used to construct a feature representation of the target object based on the historical answer records and the knowledge point information; and input the feature representation into the cognitive diagnostic model so that the cognitive diagnostic model outputs candidate exercises. The acquisition module is further configured to acquire target parameters, which are weight parameters of the cognitive diagnostic model. The weight parameters include the knowledge relevance between the knowledge points involved in the exercises, the knowledge proficiency of the target object in the knowledge points involved in the exercises, the difficulty of the exercises for the knowledge points of the target object, and the knowledge point differentiation between the knowledge points involved in the exercises. The processing module is further configured to construct a multi-objective optimization problem and constraints on the multi-objective optimization problem based on the target parameters. The constraints include at least the target object's knowledge proficiency in the knowledge points involved in the recommended exercises and the number of knowledge points involved in the recommended exercises. The objectives of the multi-objective optimization problem include at least the difficulty of the knowledge points in the recommended exercises reaching a first threshold and the knowledge relevance between the knowledge points involved in the recommended exercises and the knowledge points involved in the X exercises reaching a second threshold. An output module is configured to solve the multi-objective optimization problem based on the at least two constraints, so as to obtain the recommended problem from the candidate problems; Specifically, the processing module is used to train an initial cognitive diagnostic model based on the feature representation to obtain the cognitive diagnostic model. The weight parameters of the initial cognitive diagnostic model include initial knowledge proficiency, initial knowledge relevance, initial knowledge point difficulty, and initial knowledge point discrimination. The feature representation is input into the cognitive diagnostic model to obtain the predicted score for each exercise in the exercise bank; The candidate exercises are output from the exercise bank based on the predicted scores.

13. A computer device, characterized in that, include: Memory, processor, and bus system; The memory is used to store programs; The processor is configured to execute a program in the memory, and the processor is configured to execute the method of any one of claims 1 to 11 according to instructions in the program code; The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

14. A computer-readable storage medium comprising instructions that, when executed on a computer, cause the computer to perform the method as claimed in any one of claims 1 to 11.

Images