CN114493944A

CN114493944A - Method, device and equipment for determining learning path and storage medium

Info

Publication number: CN114493944A
Application number: CN202210051135.5A
Authority: CN
Inventors: 丁亮; 王士进; 苏喻; 沙晶
Original assignee: iFlytek Co Ltd
Current assignee: iFlytek Co Ltd
Priority date: 2022-01-17
Filing date: 2022-01-17
Publication date: 2022-05-13

Abstract

The embodiment of the application provides a learning path determining method and device, computing equipment and a storage medium. The method comprises the following steps: inputting the acquired first learning situation data of the target user into a vulnerability diagnosis model to obtain a vulnerability knowledge point of the target user; sequencing weak knowledge points of the target user to obtain weak knowledge point sequencing of the target user; and determining the learning path of the target user to the weak knowledge points according to the weak knowledge point sequence. According to the method and the device, the first learning situation data of the target user is processed through the weak point diagnosis model, accurate diagnosis of weak knowledge points of the target user can be achieved, the diagnosed weak knowledge points are sequenced, weak knowledge points which are solved preferentially and weak knowledge points which are solved later are identified, efficient learning paths are adapted for the target user, and learning efficiency of the target user is improved.

Description

Method, device and equipment for determining learning path and storage medium

Technical Field

The embodiment of the application relates to the technical field of computers, in particular to a method, a device, equipment and a storage medium for determining a learning path.

Background

The reason that the education is highly prized in the education field is to realize that weak knowledge points of students need to be found in the education field, and the weak knowledge points are subjected to reinforcement learning, so that the learning effect of the students is improved while the burden of the students is not increased.

With the increasing popularization of online learning scenes, various products combining cognitive intelligence and traditional education are applied to the ground. For example, students can be helped to recommend homework, test wrong questions and push similar questions, or weak points of the students can be judged by counting wrong question knowledge points, so as to push practice problems.

However, the current method does not accurately diagnose weak knowledge points of students, and cannot further plan a reasonable learning path according to individual conditions of the students, thereby causing the problem of low learning efficiency.

Disclosure of Invention

The embodiment of the application provides a method, a device, equipment and a storage medium for determining a learning path so as to improve the learning efficiency of students.

In a first aspect, an embodiment of the present application provides a method for determining a learned path, including:

inputting the acquired first learning situation data of the target user into a vulnerability diagnosis model to obtain a vulnerability knowledge point of the target user;

sequencing the weak knowledge points of the target user to obtain weak knowledge point sequencing of the target user;

and determining the learning path of the target user to the weak knowledge points according to the weak knowledge point sequence.

In a second aspect, an embodiment of the present application provides a learning path determining apparatus, including:

the diagnosis unit is used for inputting the acquired first learning situation data of the target user into a weak point diagnosis model to obtain weak knowledge points of the target user;

the sequencing unit is used for sequencing the weak knowledge points of the target user to obtain weak knowledge point sequencing of the target user;

and the processing unit is used for determining the learning path of the target user to the weak knowledge points according to the weak knowledge point sequencing.

In a third aspect, a computing device is provided that includes a processor and a memory. The memory is configured to store a computer program, and the processor is configured to call and execute the computer program stored in the memory to perform the method in the first aspect or each implementation manner thereof.

In a fourth aspect, a chip is provided for implementing the method in any one of the first to second aspects or implementations thereof. Specifically, the chip includes: a processor, configured to call and run a computer program from a memory, so that a device on which the chip is installed performs the method according to any one of the above first aspects or the implementation manners thereof.

In a fifth aspect, a computer-readable storage medium is provided for storing a computer program, the computer program causing a computer to perform the method of any one of the above aspects or implementations thereof.

A sixth aspect provides a computer program product comprising computer program instructions for causing a computer to perform the method of any of the above aspects or implementations thereof.

In a seventh aspect, a computer program is provided, which, when run on a computer, causes the computer to perform the method of any one of the above first aspects or implementations thereof.

In conclusion, the acquired first learning situation data of the target user are input into the vulnerability diagnosis model, so that the weak knowledge points of the target user are obtained; sequencing weak knowledge points of the target user to obtain weak knowledge point sequencing of the target user; and determining the learning path of the target user to the weak knowledge points according to the weak knowledge point sequence. In other words, according to the embodiment of the application, the first learning situation data of the target user is processed through the weak point diagnosis model, so that the weak knowledge points of the target user can be accurately diagnosed, the diagnosed weak knowledge points are sequenced, the weak knowledge points which are preferentially solved and the weak knowledge points which are later solved are identified, an efficient learning path is adapted to the target user, and the learning efficiency of the target user is improved.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and together with the description, serve to explain the principles of the application.

Fig. 1 is an application scenario diagram of a learning path determining method applied to a terminal device according to an embodiment of the present application;

fig. 2 is a schematic application scenario diagram of a learning path determining method applied to a server according to an embodiment of the present application;

fig. 3 is a schematic flowchart of a method for determining a learning path according to an embodiment of the present disclosure;

fig. 4 is a schematic diagram of a learning path determining process according to an embodiment of the present application;

fig. 5 is a schematic diagram of a learning path determining process according to an embodiment of the present application;

fig. 6 is a schematic diagram of a learning path determining process according to an embodiment of the present application;

fig. 7 is a schematic flowchart of a method for determining a learning path according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of a learning path determining apparatus according to an embodiment of the present application;

fig. 9 is a schematic block diagram of a computing device provided by an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.

It should be understood that, in the present embodiment, "B corresponding to a" means that B is associated with a. In one implementation, B may be determined from a. It should also be understood that determining B from a does not mean determining B from a alone, but may be determined from a and/or other information.

In the description of the present application, "plurality" means two or more than two unless otherwise specified.

In addition, in order to facilitate clear description of technical solutions of the embodiments of the present application, in the embodiments of the present application, terms such as "first" and "second" are used to distinguish the same items or similar items having substantially the same functions and actions. Those skilled in the art will appreciate that the terms "first," "second," etc. do not denote any order or quantity, nor do the terms "first," "second," etc. denote any order or importance.

The technical scheme provided by the application is mainly applied to the technical field of education.

Along with the implementation of the national policy of 'double reduction', the total amount and duration of the operation are required to be controlled in various places. Although the homework is an important link of the education and teaching management work of schools and is necessary supplement of classroom teaching activities, the homework of some schools has too much number, low quality and different functions, not only can not achieve the effect of knowing the temperature but also occupies the normal exercise, rest and entertainment time of students. Therefore, in recent years, personalized homework has become more and more loud, namely, homework release needs to directly hit weak knowledge points of students, thereby achieving the effect of getting twice the result with half the effort.

At present, online learning scenes are increasingly popularized, and tens of millions of students take lessons online, perform homework online, take examinations and the like. Moreover, more and more schools have also built learning situation databases for each student. Along with the accumulation of massive learning situation data, various products combining cognitive intelligence and traditional education are applied to the ground. On the basis of traditional teaching, the products bring more targeted supplement to the improvement of the student performance, and can improve the learning performance while relieving burden and reducing blood pressure for students.

The 'teaching according to the material' is a deep education thought of China for more than two thousand years, and the realization of the teaching according to the material is not only an education dream of human beings, but also a continuous pursuit of educating workers for hundreds of years. This application combines important technical field such as education assessment study, natural language processing, machine learning and deep learning, fully absorbs the teaching theory of "teaching according to the material on the basis of according with traditional education, has realized the complete process from the diagnosis of learning the feelings to the study path planning, really knows every student and learns the feelings length to let the student make a premium and avoid the weak in the learning process, establish interest in learning. The method is suitable for online learning scenes, and in some embodiments, learning paths of students can be adjusted in real time by combining real-time answering results of the students, so that the students can be helped to quickly solve weak points.

Through the application of the existing education market and the research on papers, the current scheme mainly comprises the following two types:

in some schemes, in the current education market, the main scene is based on school examinations, and similar question retrieval is carried out by using examination wrong questions of each student so as to carry out consolidation and rehearsal. The scheme is widely applied to schools at present, and the products are very popular because the method is very consistent with the thought of strengthening and re-training wrong questions in education, and teachers and students can leave the time for finding the wrong questions to quickly strengthen the wrong questions. However, the main defects of the method are commonly recognized, the method cannot be recommended without examination, the method cannot find examination questions which are not related to the examination questions, and particularly cannot be recommended to the point with lower examination evaluation. Therefore, the scheme strongly depends on the examination, so that students cannot overcome the weak points not involved in the examination, and the rollback strengthening effect cannot be achieved even if the weak points solved at the time are not covered in the next examination. In addition, the study path planning cannot be adapted in the question-based question-pushing mode, and the questions suitable for the difficulty of the students cannot be recommended according to the student study conditions, so that the recommended questions do not play a role in solving weak points, and the students are alarmed to answer the questions.

In other schemes, some tutors can establish a related knowledge point system by themselves, and according to the practice condition of students in the tutors, related questions are summarized to related knowledge points, rough probability statistics is carried out, and partial weak point diagnosis effects are also achieved, so that the tutors are assisted to carry out experience analysis, and relevant teaching plans are further carried out by the tutors according to the knowledge points with poor mastering degree. However, the scheme still stays in an auxiliary teaching link and does not derive students to help students to independently practice overcoming weak points. Briefly, assuming that the accuracy of the statistical weak point is correct, the teacher knows the weak point and needs to assign personalized homework to each student, correct each individual in time, and adjust the practice/give a lecture according to the response. Although the teaching principle of' teaching according to the situation is completely met, the burden of actual operation on teachers is very large, and the learning path cannot be automatically pushed to students or teachers.

Therefore, most of the products currently only help students to recommend homework/examination wrong questions and push similar questions (such as a personalized chemistry exercise manual), or help students to judge weak points by counting wrong-question knowledge points, so as to push some exercise questions. In essence, the weak points of the students are not accurately diagnosed according to the complete study situation and reasonable study paths are further planned according to the individual conditions of the students, so that each student can effectively solve the weak points in a limited time.

Aiming at the limitation of the existing scheme, the method for determining the learning path is different from the traditional scheme for counting from examination questions to knowledge points, and the method for determining the learning path trains an accurate weak point diagnosis model by using tens of millions of complete student learning condition data, and the model can quickly construct a learning condition picture and a knowledge mastering condition (namely a knowledge map) of each student, so that the students can be helped to accurately find weak links. And then, performing personalized sequencing on the weak points of each student, and identifying preferential solution points and delayed solution points. And finally, the learning path of the student is determined based on the sequenced weak knowledge points, so that the reasonability of the determined learning path is ensured, and the learning efficiency of the student is improved.

In order to facilitate understanding of technical solutions provided by the embodiments of the present application, before describing detailed solutions of the embodiments of the present application, an application scenario provided by the embodiments of the present application is exemplarily described with reference to fig. 1 and fig. 2 below.

Fig. 1 is an application scenario diagram of a learning path determining method applied to a terminal device according to an embodiment of the present application. In the application scenario shown in fig. 1, the terminal device 102 executes the learned path method provided in the embodiment of the present application. Specifically, as shown in fig. 1, when a user 101 triggers a learning path determination request on a terminal device 102, the terminal device 102 receives the learning path determination request, and obtains a learning path of a target user for a weak knowledge point by executing the method provided by the embodiment of the present application. Specifically, the user 101 sends the first emotion data of the target user to the terminal device 102, or the user 101 sends a learning path request to the terminal device 102, where the learning path request includes identification information of the target user, and the terminal device 102 acquires the first emotion data of the target user from a server or another platform according to the identification information of the target user. Then, the terminal device 102 inputs the obtained first learning situation data of the target user into the vulnerability diagnosis model to obtain the weak knowledge points of the target user, and determines the learning path of the target user for the weak points only, namely the weak points, according to the weak knowledge points.

In addition, the training process of each model according to the embodiment of the present application is also implemented by the terminal device 102.

Optionally, the terminal device 102 may save the trained vulnerability diagnosis model to its own storage space.

Optionally, the terminal device 102 may also store the trained vulnerability diagnostic model in a server corresponding to the terminal device 102, and when the method of the embodiment of the present application is executed, read the vulnerability diagnostic model from the server.

Fig. 2 is an application scenario diagram of the learning path determining method applied to the server according to the embodiment of the present application. In the application scenario shown in fig. 2, when a user 201 triggers a learning path determination request on a terminal device 202, the terminal device 202 receives the learning path determination request, and forwards the learning path determination request to a server 203, so that the server 203 obtains a learning path of a target user for a weak knowledge point by executing the method provided in the embodiment of the present application.

The training process of each model according to the embodiment of the present application is also realized by the server 203.

Optionally, the server 203 may save the trained vulnerability diagnosis model to its own storage space.

Optionally, the server 203 may also save the trained vulnerability diagnostic model to other devices, for example, in a distributed storage system, and when executing the method of the embodiment of the present application, the server 203 reads the vulnerability diagnostic model from the distributed storage system.

As mentioned above, the terminal device 102 and the terminal device 202 may include a mobile phone, a tablet computer, a notebook computer, a palm computer, a Mobile Internet Device (MID) or other terminal devices with data processing functions.

The server 203 may be a rack server, a blade server, a tower server, or a rack server. The server may be an independent server or a server cluster composed of a plurality of servers.

In this embodiment, the terminal apparatus 202 is connected to the server 203 via a network. The network may be a wireless or wired network such as an Intranet (Intranet), the Internet (Internet), a Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), a 4G network, a 5G network, Bluetooth (Bluetooth), Wi-Fi, or a communication network.

It should be noted that the method for determining a learning path provided in the embodiment of the present application can be applied to not only the application scenarios shown in fig. 1 or fig. 2, but also other application scenarios in which a learning path needs to be determined, and the embodiment of the present application is not particularly limited to this.

The technical solutions of the embodiments of the present application are described in detail below with reference to some embodiments. The following several embodiments may be combined with each other and may not be described in detail in some embodiments for the same or similar concepts or processes.

Fig. 3 is a schematic flow chart of a method for determining a learned path according to an embodiment of the present application, as shown in fig. 3, including:

s301, inputting the acquired first learning situation data of the target user into the vulnerability diagnosis model to obtain a vulnerability knowledge point of the target user.

The execution subject of the embodiment of the present application may be the terminal device 102 shown in fig. 1, or may be the server 203 shown in fig. 2. Optionally, the execution main body in the embodiment of the present application may also be a computing device with other specific data processing functions.

For convenience of description, the following embodiments take an execution subject as an example of a computing device.

For convenience of description, the emotion data of the target user is recorded as first emotion data.

Illustratively, the first school things data of the target user includes an answer record including the test question and the score of the test question in at least one of week examination paper, month examination paper, term examination paper, end-of-term examination paper, joint examination paper, assignment and classroom data of the school in a preset time (for example, a school term).

It should be noted that the first literary expression data of the target user, which is acquired in the embodiment of the present application, may be understood as full-quantitative literary expression data, that is, obtaining as many literary expression data of the target user as possible.

The vulnerability diagnosis model can be used for realizing the description of the learning situation image of the target user based on the first learning situation data of the target user and quickly positioning the weak knowledge points of the target user. Compared with the traditional statistical method, the scheme for determining weak knowledge points through the weak point diagnosis model has the greatest advantages that the obvious weak knowledge points (namely knowledge points related to common mistakes) of the student individuals can be reflected, correlation analysis of the knowledge points can be performed through establishing correlation among the knowledge points, and even early warning can be performed on the weak points which may become in the future.

The specific Network structure of the vulnerability diagnosis model is not limited in the embodiment of the present application, and for example, the vulnerability diagnosis model may be a timing model such as an RNN (Recurrent Neural Network), an end-to-end transform (a machine translation model), or the like.

The weak point diagnosis model is obtained through massive mathematical situation data training. Some prior tutoring institutions/education companies adopt a self-built learning knowledge system and a statistical method to realize the diagnosis of weak points of students and assist teachers in teaching. The biggest problem is that few institutions can acquire complete session data of school students, especially important examination data. Because the examination is in a serious scene, and the examination questions have high reliability and good discrimination, and other reasonable indexes, the examination is essential for the student to write the academic situation. According to the embodiment of the application, massive academic situation data of students are obtained based on the online databases such as personalized hand-screening data and accurate teaching test paper, the quality and the quantity of the data are guaranteed, and therefore when the massive academic situation data are used for training the vulnerability diagnosis model, the vulnerability diagnosis model can be fully trained, and the prediction accuracy of the vulnerability diagnosis model is guaranteed.

In the embodiment of the application, the vulnerability diagnosis model is used for diagnosing the vulnerability knowledge points of the user, wherein specific output contents of the vulnerability diagnosis model can be set.

In case 1, the vulnerability diagnosis model of the embodiment of the present application may directly output the vulnerability knowledge points of the target user, for example, directly output: knowledge points A, knowledge points B, knowledge points C and the like, so that weak knowledge points of the target user can be determined to be the knowledge points A, the knowledge points B, the knowledge points C and the like. That is to say, the embodiment of the present application may only enable the vulnerability diagnostic model to output at least one knowledge point that the target user has a poor understanding condition by limiting the output of the vulnerability diagnostic model.

In case 1, the vulnerability diagnostic model may output the grasping condition of each vulnerability knowledge point, which may be expressed by a single scalar, and may indicate the grasping condition of the vulnerability knowledge point, i.e., the degree of vulnerability, by the target user, in addition to directly outputting the vulnerability knowledge point of the target user. Therefore, when recommending the subsequent learning materials, more learning materials can be recommended for the knowledge points with larger weakness degree for reinforcement learning, and less learning materials can be recommended for the knowledge points with smaller weakness degree for reinforcement learning.

In case 2, the vulnerability diagnosis model of the embodiment of the present application outputs the grasping conditions of the target user with respect to a plurality of preset knowledge points, for example, the vulnerability diagnosis model outputs the grasping conditions of the target user with respect to each of 100 preset knowledge points. For example, the grasping condition of a knowledge point can be represented by a scalar, for example, a larger scalar corresponding to the knowledge point represents that the grasping condition of the knowledge point is better, i.e., the knowledge point is not a weak knowledge point, and a smaller scalar corresponding to the knowledge point represents that the grasping condition of the knowledge point is worse, i.e., the knowledge point is a weak knowledge point. Optionally, the scalar may also be used to indicate a weak knowledge point, for example, a larger value of the scalar indicates that the knowledge point is weaker for the target user, and a smaller value of the scalar indicates that the knowledge point is better grasped by the target user and is not a weak knowledge point.

In this case 2, the grasping conditions of the target user about each knowledge point output by the vulnerability diagnostic model may be screened, and the knowledge points whose grasping conditions are smaller than a certain preset value may be determined as the vulnerability of the target user.

Based on the network structure of the vulnerability diagnosis model, the method for obtaining the vulnerability knowledge points of the target user through the vulnerability diagnosis model in S301 includes, but is not limited to, the following methods:

the first method is that the first learning situation data of the target user is directly input into the vulnerability diagnosis model, and the vulnerability diagnosis model outputs weak knowledge points of the target user or outputs the mastering condition of the target user about preset knowledge points based on the first learning situation data of the target user. That is, in the first mode, the first school situation data of the target user is not preprocessed before being input into the vulnerability diagnosis model, but is directly input into the vulnerability diagnosis model.

In the second mode, in order to reduce the data processing amount of the vulnerability diagnostic model, the first condition data of the target user is processed, for example, converted into vector representation, before being input into the vulnerability diagnostic model. And then, inputting the processed data into the vulnerability diagnosis model, so that the data processing amount of the vulnerability diagnosis model is reduced, and the speed of determining the weak knowledge points of the target user is further improved.

In the second mode, as shown in fig. 4, the step S301 includes the following steps S301-a1 to S301-A3:

S301-A1, determining test question vector representation and score vector representation of each test question in the first emotion learning data.

The ways of determining the test question vector representation and the score vector representation of each test question in the first emotion data in S301-a1 include, but are not limited to, the following examples:

example 1, the first literary condition data of the target user includes a plurality of test questions, and a score of the target user with respect to each of the test questions. Based on the data, the first learning situation data of the target user can be input into a pre-trained neural network model A, and the neural network model A can simultaneously output the test question vector representation of the test question and the score vector representation of the user about the test question. In this way, the neural network model a can output a test question vector representation of each test question in the first emotion learning data, and a score vector representation of the target user on each test question. That is, in this example 1, the test question vector representation and the score vector representation can be obtained simultaneously by one neural network model a.

Example 2, inputting each test question in the first emotion learning data into the test question representation model to obtain test question vector representation of each test question; and processing the score of the target user about each test question in the first emotion data to obtain the score vector representation of each test question.

The test question Representation model may be a Text CNN (Text Convolutional Neural Network) type model, or a currently popular corpus pre-training model, such as BERT (Bidirectional Encoder Representation based on a converter) and ELMo (emotion from Language Models), and the embodiment of the present application does not limit the specific type of the test question Representation model.

That is, in this example 2, the determination process of the test question vector representation and the determination process of the score vector representation are two processes, in which the test question vector representation is obtained by extracting the characteristic information of the test question through a neural network model, such as a test question representation model. Since the score is a numerical scalar, the score vector representation can be obtained by performing data processing on the score.

For example, the score is directly expressed as a score vector.

For another example, the score of the target user about each test question in the first emotion learning data is normalized to obtain the score vector representation of each test question.

The embodiment of the present application does not limit the way of normalizing the scores.

In an example, a zero-mean normalization method may be adopted to normalize the score of each test question in the first emotion data of the target user, so as to obtain a score vector representation of each test question. For example, a certain test question score vector representation is obtained through the following formula (1):

wherein x is a score vector representation of the target for the test question i, xi is a score of the target user for the test question i, and σ is a standard deviation of the scores of the target user for all the test questions in the first mood data.

In another example, a min-max normalization manner may be adopted to normalize the score of each test question in the first emotion data of the target user, so as to obtain a score vector representation of each test question. For example, a certain test question score vector representation is obtained by the following formula (2):

x*＝xi-x_min/x_max-x_min (2)

wherein x is the score vector representation of the target for the test question i, xi is the score of the target user for the test question i, and x_minAnd x_maxThe minimum score value and the maximum score value of the scores of the target users about all the test questions in the first learning situation data are respectively obtained.

Optionally, other normalization methods may be used to normalize the score to obtain a score vector, where the score vector is represented as a numerical value.

S301-A2, integrating the test question vector representation and the score vector representation of all the test questions in the first emotion learning data to obtain integrated vector representation.

According to the steps, after test question vector representation and score vector representation of each test question are obtained, the test question score vector representation and the score vector representation are integrated to obtain integrated vector representation.

It should be noted that the integrated vector is represented as a single vector, that is, the test question vector representation and the score vector representation of all the test questions in the first emotion learning data are integrated into a single vector representation.

In some embodiments, since the learning process of knowledge points is sequential, for example, learning knowledge point a first and then learning knowledge point B based on knowledge point a is learned. Therefore, in order to improve the diagnosis accuracy of the weak point diagnosis model on the weak knowledge points, in the embodiment of the application, the test question vector representations and the score vector representations of all the test questions in the first learning context data are integrated according to the learning sequence of the knowledge points corresponding to the test questions in the first learning context data, so as to obtain the integrated vector representation. For example, if the learning order of the knowledge points corresponding to the test question is advanced, the test question vector representation and the score vector representation of the test question are integrated.

The embodiment of the present application does not limit the integration manner of the test question vector representation and the score vector representation.

In a possible implementation mode 1, the test question vector representation and the score vector representation of each test question are spliced and then spliced to obtain the integrated vector representation. For example, the first emotion learning data of the target user includes n test questions, wherein a test question vector of a test question i is represented as a test question vector representation i, a score vector of the test question i is represented as a score vector representation i, and the integration vector obtained by splicing according to the method is represented as: { test question vector represents 1, score vector represents 1, test question vector represents 2, score vector represents 2, …, test question vector represents i, score vector represents i, …, test question vector represents n, and score vector represents n }.

In a possible implementation manner 2, the test question vector representation and the score vector representation of each test question are subjected to point multiplication to obtain a point multiplication result corresponding to each test question, and the point multiplication results corresponding to all the test questions in the first learning situation data are spliced to obtain an integrated vector representation.

For example, the first emotion data of the target user includes n test questions, where a test question vector of a test question i is represented as a test question vector representation i, a score vector of the test question i is represented as a score vector representation i, the test question vector representation i and the score vector representation i of the test question i are dot-multiplied, and since the score vector representation i is a numerical value, a dot-multiplication result of dot-multiplying the test question vector representation i and the score vector representation i is a result of multiplying each dimension of the test question vector representation i by the score vector representation i. Assuming that the test question vector represents i as a 10-dimensional vector and the score vector represents i as 0.2, the point multiplication result of the test question vector representation i and the score vector representation i is that each dimension of 10 dimensions of the test question vector representation i is multiplied by 0.2 to obtain a 10-dimensional vector. And executing the step of the test question i on each test question in the n test questions according to the method to obtain a dot product result corresponding to each test question. Then, the dot product results corresponding to the n test questions are spliced to obtain an integrated vector representation, for example, the dimension of the integrated vector representation is n × 10.

In a possible implementation manner 3, the test question vector representation and the score vector representation of each test question are subjected to point multiplication to obtain a point multiplication result corresponding to each test question, and the point multiplication results corresponding to all the test questions in the first learning situation data are averaged in each dimension to obtain an integrated vector representation.

In the embodiment of the present application, the dot product result is also a vector, the number of elements included in the vector is referred to as a dimension, for example, if the vector includes 10 elements, the dimension of the vector is determined to be 10.

Continuing with the example that the first emotion data of the target user includes n test questions, where a test question vector of the test question i is represented as a test question vector representation i, a score vector of the test question i is represented as a score vector representation i, and the test question vector representation i and the score vector representation i of the test question i are dot-multiplied, and since the score vector representation i is a single numerical value, a dot-multiplied result obtained by dot-multiplying the test question vector representation i and the score vector representation i is represented as a score vector representation i by multiplying each dimension of the test question vector representation i. Assuming that the test question vector represents i as a 10-dimensional vector and the score vector represents i as 0.2, the point multiplication result of the test question vector representation i and the score vector representation i is that each dimension of 10 dimensions of the test question vector representation i is multiplied by 0.2 to obtain a 10-dimensional vector. And executing the step of the test question i on each test question in the n test questions according to the method to obtain a dot product result corresponding to each test question. Then, the dot product results corresponding to the n test questions are averaged in each dimension, for example, the dimension of the dot product result corresponding to each test question is 10, for each dimension of the 10 dimensions, the dot product results corresponding to the n test questions are averaged in the dimension, and the average value is used as a value integrated by the dimension, so that an integrated 10-dimensional vector representation can be obtained.

S301-A3, inputting the integrated vector representation into a weak point diagnosis model to obtain weak knowledge points of the target user.

According to the method, the integrated vector representation of the test question vector representation and the score vector representation is obtained, and then the integrated vector representation is input into the weak point diagnosis model to obtain the weak knowledge points of the target user.

In some embodiments, if it is necessary to acquire the knowledge of the target user about the specific knowledge point, the step S301 includes the following steps:

S301-B, inputting the first learning situation data of the target user and the preset knowledge point range into the vulnerability diagnosis model to obtain the weak knowledge points of the target user.

According to the embodiment of the application, the accurate diagnosis of the weak knowledge points of the target user is realized through the weak point diagnosis model based on the first learning situation data of the target user. Next, the following steps S302 and S303 are performed to determine the learned route of the target user.

S302, sequencing weak knowledge points of the target user to obtain weak knowledge point sequencing of the target user.

According to the above, the weak knowledge points of the target user are accurately found, but the weak knowledge points are also weakly sorted, namely the weak knowledge points have a certain learning sequence for different students, and the learning effect is directly influenced by the sequence theory.

Based on this, after weak knowledge points of the target user are obtained, weak knowledge points of the target user are sequenced to obtain weak knowledge point sequencing of the target user, and a learning path is determined based on the weak knowledge points sequenced by the target user.

The method for sorting the weak knowledge points of the target user in S302 includes, but is not limited to, the following:

in the first way, the criticality of all knowledge points is different in the process of learning by students, and some knowledge points are used as key knowledge points, which need to be mastered by each student and are relatively large in examinations. Some knowledge points are not key knowledge points, such as extended knowledge points, do not need to be mastered by each student, and are smaller in the examination. Therefore, in the first mode, the weak knowledge points of the target user can be sequenced according to whether the weak knowledge points of the target user are key knowledge points or not, so that the target user can solve the key knowledge points preferentially and solve the non-key knowledge points later, and the learning efficiency of students is improved.

Specifically, the key knowledge points are obtained, where the key knowledge points may be obtained from a teaching outline, or from teaching materials of teachers, or from historical test papers, and for example, the key knowledge points with frequent occurrence and large score in the test papers are determined as the key knowledge points.

And then, taking the weak knowledge points belonging to the key knowledge points in the weak knowledge points of the target user as weak knowledge points to be solved preferentially, and arranging the weak knowledge points at the top. And (3) taking the weak knowledge points which do not belong to the key knowledge points in the weak knowledge points of the target user as the weak knowledge points to be solved in a delayed manner, and arranging the weak knowledge points behind the weak knowledge points which are solved in a priority manner, thereby obtaining the weak knowledge point sequencing of the target user.

And secondly, sequencing the weak knowledge points of the target user according to the grasping condition of the weak knowledge points of the users close to the achievement of the target user through the recent development area theory. Specifically, the step S302 includes the following steps S302-A1 to S302-A3:

S302-A1, obtaining the examination ranking of the target user, and determining at least one user in the examination ranking floating preset interval of the target user as a sample user of the target user;

S302-A2, comparing the weak knowledge points of the target user with the weak knowledge points of the target user, determining knowledge points mastered by the target user in the weak knowledge points of the target user as weak knowledge points which are preferentially solved by the target user, and determining knowledge points not mastered by the target user in the weak knowledge points of the target user as weak knowledge points which are solved by the target user after delay;

and S302-A3, obtaining weak knowledge point sequencing of the target user according to weak knowledge points which are solved by the target user preferentially and weak knowledge points which are solved later.

In actual learning situations, the students with poor learning performance and the students with good learning performance have different degrees of acceptance for the same knowledge point, for example, the knowledge points "analysis geometry is the comprehensive application of conic curve" are weak knowledge points for both the student a with poor learning performance and the student B with good learning performance, but when the student B recommends practice materials for the knowledge points, the student B is easy to accept and learn. However, for the student a, the weak knowledge point is a panic area, and the expected learning effect cannot be achieved even if the student a spends more time, because the student a does not know the relevant basic knowledge, directly learns the more comprehensive weak knowledge point, and has a great impact on the mind of the student a, and the student a may be unwilling to learn.

In order to solve the technical problem, the weak knowledge points of the target user are sorted by adopting a recent development area theory. Specifically, the examination rank of the target user is obtained, and the examination rank may be the latest examination rank, the rank of the target user is raised to a preset interval, for example, the rank of the target user is 30 th, the raised preset interval is 10%, the obtained interval of the target user is the users between 20 th and 29 th, and at least one user between 20 th and 29 th is taken as the target user's list user. That is to say, in the embodiment of the application, the first students close to the target user in the examination ranking are used as the list-like users of the target user, and the method has the advantage of always comparing the user with a slightly stronger standard than the user. The weak knowledge points of the target user are compared with the weak knowledge points of the target user, the knowledge points mastered by the target user in the weak knowledge points of the target user are determined as the weak knowledge points which are preferentially solved by the target user, the knowledge points which are not mastered by the target user in the weak knowledge points of the target user are determined as the weak knowledge points or parts which are solved by the target user after delay, and therefore a stepped learning effect is achieved, the target user is prevented from directly falling into a learning panic area, and the learning interest and the learning efficiency of the target user are improved.

In some embodiments, weak knowledge points that are preferentially solved by the target user may be directly arranged before weak knowledge points that are solved later, so as to obtain weak knowledge point ranking of the target user.

In some embodiments, different knowledge points are difficult to overcome in certain areas due to regional differences in education itself, but easy in other areas due to differences in the manner of investigation or education. Therefore, in order to further improve the accuracy of sorting the weak knowledge points of the target user, the weak knowledge points of the target user determined according to the method need to be adjusted in combination with the weak knowledge points of the target area where the target user is located. Specifically, as shown in FIG. 5, the step S302-A3 includes the following steps S302-A31-S302-A33:

and S302-A31, obtaining the initial sequencing of the weak knowledge points of the target user according to the weak knowledge points which are solved by the target user preferentially and the weak knowledge points which are solved later.

For example, the weak knowledge points that are solved preferentially by the target user are arranged before the weak knowledge points that are solved later, and the initial sequencing of the weak knowledge points of the target user is obtained.

S302-A32, second learning situation data of a target area where the target user is located are obtained, and weak knowledge points and weights of the weak knowledge points of the target area are obtained based on the second learning situation data.

For convenience of description, the emotion data of the target area where the target user is located is recorded as the second emotion data.

The embodiment of the application does not limit the way of obtaining the weak knowledge points of the target area based on the second learning situation data.

For example, the weak knowledge points of students in the target area are obtained through statistics.

The weight of the weak knowledge point may be a preset value set according to whether the weak knowledge point is a key knowledge point. For example, if the weak knowledge point a is a key knowledge point, a larger weight may be set, and if the weak knowledge point B is a non-key knowledge point, a smaller weight may be set.

S302-A33, adjusting the initial sequencing of the weak knowledge points of the target user according to the weak knowledge points of the target area and the weights of the weak knowledge points to obtain the sequencing of the weak knowledge points of the target user.

Specifically, according to the weak knowledge points of the target area and the weights of the weak knowledge points, the initial sequencing of the weak knowledge points of the target user is finely adjusted in a weighting mode, and the sequencing of the weak knowledge points of the target user is obtained.

For example, assume that the initial ordering of weak knowledge points for the target user is: the system comprises a knowledge point A, a knowledge point B and a knowledge point C, wherein the mastery degree of the board-like user on the knowledge point A is 100%, the mastery degree of the board-like user on the knowledge point B is 90%, and the mastery degree of the board-like user on the knowledge point C is 50%. According to the weak knowledge points and the weights of the weak knowledge points of the target area, in the target area, the weight of the knowledge point A is 1, the weight of the knowledge point B is 4, and the weight of the knowledge point C is 1.2, so that the mastery degree and the weight of each knowledge point are multiplied, and the weak knowledge points of the target user are ranked according to the size of the multiplied result from large to small, so that the weak knowledge point ranking of the target user is obtained, namely the weak knowledge point ranking of the target user is as follows: knowledge point B, knowledge point A and knowledge point C.

According to the embodiment of the application, the weak point sequencing of the target users is realized by combining the second learning situation data of the target area where the target users are located, and when weak knowledge point learning is carried out according to the weak point sequencing, the learning enthusiasm of the target users and the learning of key knowledge points can be improved.

And S303, determining the learning path of the target user to the weak knowledge points according to the weak knowledge point sequencing.

In the embodiment of the application, the learning path of the target user for the weak knowledge points can be understood as recommending the learning materials corresponding to the weak knowledge points for the target user according to the sequence of the weak knowledge points. For example, the target user's weak knowledge points are ordered as: knowledge point B, knowledge point A and knowledge point C, the learning path is as follows: recommending the relevant learning materials of the knowledge point B, recommending the relevant learning materials of the knowledge point A after the learning of the relevant materials of the knowledge point B is finished, and recommending the relevant learning materials of the knowledge point C after the learning of the relevant materials of the knowledge point B is finished.

The implementation manners of S303 include, but are not limited to, the following:

in the first mode, according to the weak knowledge point sequencing, the learning data corresponding to each weak knowledge point is inquired one by one in the existing learning resource library to obtain the learning path of the target user to the weak knowledge points. Specifically, the learning resources corresponding to the first weak knowledge point in the weak knowledge point sequence are recommended to the target user, then the learning resources corresponding to the second weak knowledge point in the weak knowledge point sequence are recommended, and the rest is done in sequence, so that the target user can learn according to the weak knowledge point sequence, a stepped learning effect is achieved, the target user cannot directly fall into a learning panic area, and the learning efficiency is improved.

And in the second mode, learning resources are automatically integrated through the model, and a learning path of the target user to the weak knowledge points is generated. Namely, the above S303 includes the following steps S303-A1 to S303-A3:

and S303-A1, determining the characteristic information output by a middle layer of the weak point diagnosis model as the learning and emotion image of the target user.

For example, the feature information output by any middle layer of the vulnerability diagnosis model is determined as the emotional image of the target user.

For another example, the feature information output by the penultimate layer network of the vulnerability diagnosis model is determined as the emotional image of the target user.

In some embodiments, a literary-emotion image, which is a vector, may be referred to as a literary-emotion characterization.

S303-A2, inputting the learning emotion images and weak knowledge point sequences of the target users and learning resources of the target area where the target users are located into a path representation model to obtain the learning path representation of the target users.

The learning resources comprise exercises, videos, knowledge cards and the like.

Optionally, the learning material further includes a label, and the label may be a difficulty label, a knowledge point label, a region attribute label, or the like.

In some embodiments, the learning situation image of the target user, the weak knowledge point sequence and the learning resources of the target area where the target user is located are directly input into the path representation model to obtain the learning path representation of the target user.

In some embodiments, the learning situation image of the target user, the weak knowledge point sequence and the learning resource of the target area where the target user is located are converted into vector representation, and then input into the path representation model to obtain the learning path representation of the target user. The learning-emotion image of the target user is a vector, and conversion is not needed. Specifically, weak knowledge point sequencing and learning resources of a target area where a target user is located are converted into vector representation.

Illustratively, as shown in fig. 6, weak knowledge point sorting and learning resources are mapped into weak knowledge point vectors and learning resource vectors respectively in a one-hot form. And then, inputting the learning situation portrait, the weak knowledge point vector and the learning resource vector into a path representation model to obtain the learning path representation of the target user.

The embodiment of the present application does not limit the specific network structure of the path characterization model, and optionally, the path characterization model is a shallow neural network model.

And S303-A3, obtaining the learning path of the target user to the weak knowledge points according to the learning path representation of the target user and the preset learning path representations of N users, wherein N is a positive integer.

In some embodiments, according to the learning path characterization of the target user and the preset learning path characterizations of the N users, calculating the similarity between the learning path characterization of the target user and each learning path characterization of the learning path characterizations of the N users. And determining the learning path with the highest similarity representing the corresponding learning path as the learning path of the target user to the weak knowledge point.

In some embodiments, as shown in FIG. 6, the above-mentioned S303-A3 includes the following steps S303-A31 to S303-A33:

S303-A31, determining the similarity between the learned path representation of the target user and the learned path representation of each user in the N users;

S303-A32, acquiring M users with the similarity meeting the preset condition from the N users, wherein M is a positive integer less than or equal to N.

For example, M users with similarity greater than a preset value are obtained from the N users.

For another example, the first M users with the largest similarity are obtained from the N users.

And S303-A33, determining the learning path of one user in the M users as the learning path of the target user to the weak knowledge points.

According to the embodiment of the application, the similarity between the learning path representation of the target user and the learning path representation of each user in the N users is determined, and a batch of dynamic isomorphic users, namely M users which are quite similar to the learning capacity structure of the target user, are locked from the N users. And then, according to the learning paths of the M users, determining the learning path of the target user to the weak knowledge points.

In some embodiments, the learning path of any one of the M users is determined as the learning path of the target user for the weak knowledge points.

In some embodiments, the learning path of the user with the largest similarity value among the M users is determined as the learning path of the target user for the weak knowledge point.

In some embodiments, as shown in FIG. 6, S303-A33 includes the following steps S303-A331 and S303-A332:

S303-A331, determining a target learning path representation with highest similarity and lowest cost with the learning path representations of the target users in the learning path representations of the M users;

and S303-A332, according to the corresponding learning path represented by the target learning path, determining the learning path of the target user to the weak knowledge points.

The method for determining the target learning path representation with the highest similarity and the lowest cost with the learning path representation of the target user in the learning path representations of the M users is not limited in the embodiment of the application.

Illustratively, a target learning path representation with the highest similarity and the lowest cost to the learning path representations of the target users is determined from the learning path representations of the M users by a Dynamic Time Warping (DTW) method.

For example, a target learning path representation which is most similar to the learning path representation of the target user and has the smallest cost is calculated by DTW ═ min { sqrt (Σ wk)/k, where wk ═ i, j, i is the ith element in the target learning path representation, j is the jth element in the learning path representation of one user in the M users, and k is the number of elements. The learning path corresponding to the target learning path representation can be considered to be the most suitable for the learning situation of the target user, and the most effective learning path is attacked and restrained by the weak knowledge points of the target user instead of being the shortest path singly.

In some embodiments, the target learning path is characterized by the corresponding learning path, and the learning path of the target user for the weak knowledge point is determined.

In other embodiments, when there is a real-time interaction of the students in the learning path, i.e., the students answer and recommend scenes, the selection of the learning path resources/exercises can be adjusted in real time, so that the learning path is more effective. Specifically, as shown in fig. 6, a learning path corresponding to the target learning path representation is determined as an initial learning path of the target user to the weak knowledge point; acquiring a real-time answer result of a target user, taking the real-time answer result as an incentive, and updating an initial learning path through a reinforcement learning model; and determining the updated initial learning path as the learning path of the target user to the weak knowledge points. According to the embodiment of the application, the learning path is adaptively adjusted according to the online answers of students, so that the target users have different personalized homework, and invalid repeated training is reduced.

The reinforcement learning model is obtained by training historical student learning situation data in a school as training data. That is to say, the training source of the reinforcement learning model is historical student learning situation data, and the reinforcement learning model can quickly find out a proper learning path which is most similar to the learning path state fitting of the target user and is used for quickly attacking the weak knowledge point subsequently according to the initial learning path and answering feedback of the target user by modeling tens of millions of historical student learning path processes.

In some embodiments, a real-time answer result of the target user can be obtained, the real-time answer result is mapped into an answer result vector through a language model, and then the learning situation picture, the weak knowledge point sequencing, the learning resource and the answer result vector are input into a path representation model to obtain a learning path representation of the target user. Specifically, the learning situation picture, the weak knowledge point vector, the learning resource vector and the answering result vector are input into a path representation model to obtain the learning path representation of the target user.

Optionally, the real-time answer result may be mapped to an answer result vector by the language model BERT.

The application scenarios of the embodiment of the present application include, but are not limited to, the following scenarios:

in the scene of arranging paper work in the teacher personalized work, the weak point diagnosis model can automatically position weak knowledge points of target users in a learning range and sequence the weak knowledge points by inputting first learning situation data and the current learning range of the target users. Then, according to the learning situation picture of the target user, weak knowledge point sequencing and the learning resource of the target area where the target user is located, the learning path representation of the target user is determined, according to the learning path representation of the target user, the learning path when the most similar students (including a knowledge ability structure and a learning path structure) attack the weak knowledge points of the target user is found, namely, a special training item needing to be trained is recommended for each weak knowledge point of the target user at one time, and the training question selection of each weak knowledge point is to automatically select training questions with different difficulties and different investigation surfaces in a test question bank according to the learning ability of the target user.

And a scene 2 for online exercise of students according to the recommended weak points, wherein the scene 2 is updated to the reinforcement learning model in real time as a feedback according to the real-time answering result of the target user on the basis of the scene 1, and the reinforcement learning model adaptively adjusts the current difficulty of the test questions, changes the test question points or calls related video resources/learning card resources according to the change of the learning condition, the learning level (learning excellence, learning middle learning badness and the like), and the answering condition (score and error positions) of the target user in real time, so that the weak knowledge points of the students are quickly solved.

According to the method for determining the learning path, the weak knowledge points of the target user are obtained by inputting the acquired first learning situation data of the target user into the weak point diagnosis model; sequencing weak knowledge points of the target user to obtain weak knowledge point sequencing of the target user; and determining the learning path of the target user to the weak knowledge points according to the weak knowledge point sequence. In other words, according to the embodiment of the application, the first learning situation data of the target user is processed through the weak point diagnosis model, so that the weak knowledge points of the target user can be accurately diagnosed, the diagnosed weak knowledge points are sequenced, the weak knowledge points which are preferentially solved and the weak knowledge points which are later solved are identified, an efficient learning path is adapted to the target user, and the learning efficiency of the target user is improved.

The method for determining the learning path provided by the embodiment of the present application is described above, and a training process of the vulnerability diagnosis model is described below.

Fig. 7 is a schematic flowchart of a method for determining a learned path according to an embodiment of the present application, as shown in fig. 7, including:

s401, acquiring historical situation data of K students, wherein K is a positive integer.

The historian data of the K students comprises at least one of week examination paper, month examination paper, interim examination paper, end-of-term examination paper, joint examination paper, homework and classroom data of the K students in a preset historical time period.

For example, in the embodiment of the present application, response records of student test papers of K students in the past 5 years in the week, month, date, end, and joint examination of the school are screened out, and the scores of the student test papers and the scores of each question are processed to obtain the historical situation data of the K students.

Optionally, the obtained work and classroom data can be used as an additional supplement to the historian data.

S402, inputting the history data of the K students into the weak point diagnosis model to obtain the predicted weak knowledge points of each student of the K students.

The vulnerability diagnosis model is a time sequence model, can successfully capture the association between similar questions, gathers the questions such as functions and geometries under the same concept, and describes the association between knowledge points of students at each time by combining the answering time of the test questions.

The implementation process of S402 is substantially the same as the implementation process of S301, and reference is made to the description of S301, which is not described herein again.

And S403, determining the loss between the predicted weak knowledge points of each of the K students and the real weak knowledge points of each of the K students.

In some embodiments, the actual weak knowledge point of each of the K students may be whether the student answers the test questions in the future, and maps the test questions to the corresponding knowledge point, and further, whether the knowledge point is a weak knowledge point, for example, if the test questions answer correctly, it is determined that the knowledge point corresponding to the test question is not a weak knowledge point, and if the test questions answer incorrectly, it is determined that the knowledge point corresponding to the test question is a weak knowledge point.

In some embodiments, the method of manual marking is directly adopted to mark whether students are weak at each knowledge point.

Optionally, the loss function adopted in the embodiment of the present application may be a cross entropy loss function.

And S404, training the vulnerability diagnosis model according to the loss.

According to the method, after the loss is determined, parameters in the vulnerability diagnosis model are updated by using the loss, whether the updated vulnerability diagnosis model reaches the preset training end condition or not is judged, and if the training end condition is reached, the updated vulnerability diagnosis model is used as the trained vulnerability diagnosis model to execute the method of the embodiment. And if the training does not reach the training ending condition, continuously updating the weak point diagnosis model by referring to the method until the training ending condition is reached. The condition for ending the model training can be that the prediction result of the model reaches a preset value, or the training times of the model reach the preset times.

According to the weak point diagnosis method and device, the weak point diagnosis model is trained by using the mass science situation data, sufficient training of the model is achieved, and the trained model can accurately diagnose weak knowledge points of the user.

The preferred embodiments of the present application have been described in detail with reference to the accompanying drawings, however, the present application is not limited to the details of the above embodiments, and various simple modifications can be made to the technical solution of the present application within the technical idea of the present application, and these simple modifications are all within the protection scope of the present application. For example, the various features described in the foregoing detailed description may be combined in any suitable manner without contradiction, and various combinations that may be possible are not described in this application in order to avoid unnecessary repetition. For example, various embodiments of the present application may be arbitrarily combined with each other, and the same should be considered as the disclosure of the present application as long as the concept of the present application is not violated.

It should also be understood that, in the various method embodiments of the present application, the sequence numbers of the above-mentioned processes do not imply an execution sequence, and the execution sequence of the processes should be determined by their functions and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

Method embodiments of the present application are described in detail above with reference to fig. 3 to 7, and apparatus embodiments of the present application are described in detail below with reference to fig. 8 to 9.

Fig. 8 is a schematic structural diagram of a device for determining a learning path according to an embodiment of the present application. The learning path determining apparatus may be a computing device, or may be a component (e.g., an integrated circuit, a chip, or the like) of the computing device, and the computing device may be a server shown in fig. 2 or a terminal device shown in fig. 1.

As shown in fig. 8, the learned path determining apparatus 10 includes:

the diagnosis unit 11 is configured to input the acquired first learning situation data of the target user into a vulnerability diagnosis model to obtain a vulnerability knowledge point of the target user;

a sorting unit 12, configured to sort the weak knowledge points of the target user to obtain a weak knowledge point sorting of the target user;

and the processing unit 13 is configured to determine a learning path of the target user to the weak knowledge points according to the weak knowledge point sequence.

In some embodiments, the diagnosis unit 11 is specifically configured to determine a test question vector representation and a score vector representation of each test question in the first emotional data; integrating the test question vector representation and the score vector representation of all the test questions in the first learning situation data to obtain integrated vector representation; and inputting the integrated vector representation into the vulnerability diagnosis model to obtain the vulnerability knowledge points of the target user.

In some embodiments, the diagnosing unit 11 is specifically configured to input each test question in the first emotion learning data into the test question representation model, so as to obtain a test question vector representation of each test question; and processing the score of the target user about each test question in the first learning situation data to obtain a score vector representation of each test question.

In some embodiments, the diagnosing unit 11 is specifically configured to perform normalization processing on the score of the target user with respect to each test question in the first emotion data, so as to obtain a score vector representation of each test question.

In some embodiments, the diagnosing unit 11 is specifically configured to integrate the test question vector representations and the score vector representations of all the test questions in the first learning context data according to the learning sequence of the knowledge points corresponding to the test questions in the first learning context data, so as to obtain the integrated vector representations.

In some embodiments, the diagnosing unit 11 is specifically configured to perform point multiplication on the test question vector representation and the score vector representation of each test question to obtain a point multiplication result corresponding to each test question, and splice the point multiplication results corresponding to all the test questions in the first learning context data to obtain the integrated vector representation; or, performing point multiplication on the test question vector representation and the score vector representation of each test question to obtain a point multiplication result corresponding to each test question, and averaging the point multiplication results corresponding to all the test questions in the first learning situation data in each dimension to obtain the integrated vector representation.

In some embodiments, the sorting unit 12 is specifically for

Acquiring the examination rank of the target user, and determining at least one user in a preset interval on which the examination rank of the target user floats as a sample user of the target user;

comparing the weak knowledge points of the target user with the weak knowledge points of the target user, determining knowledge points mastered by the target user in the weak knowledge points of the target user as weak knowledge points which are preferentially solved by the target user, and determining knowledge points not mastered by the target user in the weak knowledge points of the target user as weak knowledge points which are solved by the target user after delay; and obtaining weak knowledge point sequencing of the target user according to the weak knowledge points which are solved by the target user preferentially and the weak knowledge points which are solved later.

In some embodiments, the sorting unit 12 is specifically configured to obtain an initial sorting of weak knowledge points of the target user according to the weak knowledge points that are preferentially solved by the target user and the weak knowledge points that are solved later; acquiring second learning situation data of a target area where the target user is located, and acquiring weak knowledge points of the target area and weights of the weak knowledge points based on the second learning situation data; and adjusting the initial sequencing of the weak knowledge points of the target user according to the weak knowledge points of the target area and the weights of the weak knowledge points to obtain the sequencing of the weak knowledge points of the target user.

In some embodiments, the processing unit 13 is specifically configured to determine feature information output by an intermediate layer of the vulnerability diagnosis model as a mathematical expression of the target user; inputting the learning situation picture of the target user, the weak knowledge point sequence and the learning resources of the target area where the target user is located into a path representation model to obtain the learning path representation of the target user; and obtaining the learning path of the target user to the weak knowledge points according to the learning path representation of the target user and the preset learning path representations of N users, wherein N is a positive integer.

In some embodiments, the processing unit 13 is specifically configured to determine a similarity between the learned path representation of the target user and the learned path representation of each of the N users; obtaining M users with the similarity meeting a preset condition from the N users, wherein M is a positive integer less than or equal to N; and determining the learning path of one user in the M users as the learning path of the target user to the weak knowledge point.

In some embodiments, the processing unit 13 is specifically configured to determine, from the learned path characterizations of the M users, a target learned path characterization with the highest similarity and the smallest cost with the learned path characterization of the target user; and determining the learning path of the target user to the weak knowledge points according to the learning path corresponding to the target learning path characterization.

In some embodiments, the processing unit 13 is specifically configured to determine the target learning path as an initial learning path of the target user for a weak knowledge point, where the learning path corresponds to the target learning path; acquiring a real-time answering result of the target user, taking the real-time answering result as an incentive, and updating the initial learning path through a reinforcement learning model; and determining the updated initial learning path as the learning path of the target user to the weak knowledge points.

In some embodiments, the reinforcement learning model is trained by using historical student learning condition data as training data.

In some embodiments, the situational representation of the target user is feature information output by a penultimate network of the vulnerability diagnostic model.

In some embodiments, the processing unit 13 is specifically configured to map the weak knowledge point ranking and the learning resources into a weak knowledge point vector and a learning resource vector respectively in a one-hot form; and inputting the learning situation picture, the weak knowledge point vector and the learning resource vector into the path representation model to obtain the learning path representation of the target user.

In some embodiments, the processing unit 13 is specifically configured to obtain a real-time answer result of the target user, and map the real-time answer result into an answer result vector through a language model; and inputting the learning situation picture, the weak knowledge point vector, the learning resource vector and the answering result vector into the path representation model to obtain the learning path representation of the target user.

In some embodiments, the diagnosis unit 11 is specifically configured to input the first learning data of the target user and a preset knowledge point range into a vulnerability diagnosis model, so as to obtain a vulnerability of the target user.

In some embodiments, the processing unit 13 is further configured to obtain historian data of K students, where K is a positive integer; inputting the historical situation data of the K students into the weak point diagnosis model to obtain a predicted weak knowledge point of each student of the K students; determining a loss between the predicted weak knowledge points of each of the K students and the true weak knowledge points of each of the K students; and training the vulnerability diagnosis model according to the loss.

In some embodiments, the K student history data includes at least one of week examination paper, month examination paper, end examination paper, joint examination paper, assignment and classroom data of the K students over a preset historical period of time.

Optionally, the vulnerability diagnosis model is a time sequence model.

It is to be understood that apparatus embodiments and method embodiments may correspond to one another and that similar descriptions may refer to method embodiments. To avoid repetition, the description is omitted here. Specifically, the apparatus shown in fig. 7 may perform the embodiment of the method described above, and the foregoing and other operations and/or functions of each module in the apparatus are respectively for implementing the embodiment of the method corresponding to the computing device, and are not described herein again for brevity.

The apparatus of the embodiments of the present application is described above in connection with the drawings from the perspective of functional modules. It should be understood that the functional modules may be implemented by hardware, by instructions in software, or by a combination of hardware and software modules. Specifically, the steps of the method embodiments in the present application may be implemented by integrated logic circuits of hardware in a processor and/or instructions in the form of software, and the steps of the method disclosed in conjunction with the embodiments in the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. Alternatively, the software modules may be located in random access memory, flash memory, read only memory, programmable read only memory, electrically erasable programmable memory, registers, and the like, as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps in the above method embodiments in combination with hardware thereof.

Fig. 9 is a schematic block diagram of a computing device provided in an embodiment of the present application, and configured to execute the above method embodiment.

As shown in fig. 9, the computing device 30 may include:

a memory 31 and a processor 32, the memory 31 being arranged to store a computer program 33 and to transfer the program code 33 to the processor 32. In other words, the processor 32 may call and run the computer program 33 from the memory 31 to implement the method in the embodiment of the present application.

For example, the processor 32 may be adapted to perform the above-mentioned method steps according to instructions in the computer program 33.

In some embodiments of the present application, the processor 32 may include, but is not limited to:

general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, and the like.

In some embodiments of the present application, the memory 31 includes, but is not limited to:

volatile memory and/or non-volatile memory. The non-volatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable PROM (EEPROM), or a flash Memory. Volatile Memory can be Random Access Memory (RAM), which acts as external cache Memory. By way of example, but not limitation, many forms of RAM are available, such as Static random access memory (Static RAM, SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic random access memory (Synchronous DRAM, SDRAM), Double Data Rate Synchronous Dynamic random access memory (DDR SDRAM), Enhanced Synchronous SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

In some embodiments of the present application, the computer program 33 may be divided into one or more modules, which are stored in the memory 31 and executed by the processor 32 to perform the method of recording pages provided herein. The one or more modules may be a series of computer program instruction segments capable of performing certain functions, the instruction segments describing the execution of the computer program 33 in the computing device.

As shown in fig. 9, the computing device 30 may further include:

a transceiver 34, the transceiver 34 being connectable to the processor 32 or the memory 31.

The processor 32 may control the transceiver 34 to communicate with other devices, and specifically, may transmit information or data to the other devices or receive information or data transmitted by the other devices. The transceiver 34 may include a transmitter and a receiver. The transceiver 34 may further include one or more antennas.

It should be understood that the various components in the computing device 30 are connected by a bus system that includes a power bus, a control bus, and a status signal bus in addition to a data bus.

According to an aspect of the present application, there is provided a computer storage medium having a computer program stored thereon, which, when executed by a computer, enables the computer to perform the method of the above-described method embodiments. In other words, the present application also provides a computer program product containing instructions, which when executed by a computer, cause the computer to execute the method of the above method embodiments.

According to another aspect of the application, a computer program product or computer program is provided, comprising computer instructions stored in a computer readable storage medium. The computer instructions are read by a processor of the computing device from the computer-readable storage medium, and the processor executes the computer instructions to cause the computing device to perform the method of the above-described method embodiment.

In other words, when implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. The procedures or functions described in accordance with the embodiments of the present application occur, in whole or in part, when the computer program instructions are loaded and executed on a computer. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored on a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website, computer, server, or data center to another website, computer, server, or data center via wire (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that includes one or more of the available media. The usable medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a Digital Video Disk (DVD)), or a semiconductor medium (e.g., a Solid State Disk (SSD)), among others.

Those of ordinary skill in the art will appreciate that the various illustrative modules and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the module is merely a logical division, and other divisions may be realized in practice, for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or modules, and may be in an electrical, mechanical or other form.

Modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical modules, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. For example, functional modules in the embodiments of the present application may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for determining a learned route, comprising:

2. The method of claim 1, wherein the inputting the acquired first learning situation data of the target user into a vulnerability diagnosis model to obtain the vulnerability knowledge points of the target user comprises:

determining test question vector representation and score vector representation of each test question in the first learning situation data;

integrating the test question vector representation and the score vector representation of all the test questions in the first learning situation data to obtain integrated vector representation;

and inputting the integrated vector representation into the weak point diagnosis model to obtain weak knowledge points of the target user.

3. The method of claim 2, wherein determining a test question vector representation and a score vector representation for each test question in the first emotional data comprises:

inputting each test question in the first learning situation data into a test question representation model to obtain test question vector representation of each test question;

and processing the score of the target user about each test question in the first learning situation data to obtain a score vector representation of each test question.

4. The method of claim 2, wherein the integrating the test question vector representations and the score vector representations of all the test questions in the first emotional data to obtain an integrated vector representation comprises:

and integrating the test question vector representation and the score vector representation of all the test questions in the first learning situation data according to the learning sequence of the knowledge points corresponding to the test questions in the first learning situation data to obtain the integrated vector representation.

5. The method according to any one of claims 2 to 4, wherein the integrating the test question vector representations and the score vector representations of all the test questions in the first emotional data to obtain an integrated vector representation comprises:

performing point multiplication on the test question vector representation and the score vector representation of each test question to obtain a point multiplication result corresponding to each test question, and splicing the point multiplication results corresponding to all the test questions in the first learning situation data to obtain the integrated vector representation; alternatively, the first and second electrodes may be,

and performing point multiplication on the test question vector representation and the score vector representation of each test question to obtain a point multiplication result corresponding to each test question, and averaging the point multiplication results corresponding to all the test questions in the first learning situation data in each dimension to obtain the integrated vector representation.

6. The method according to any one of claims 1 to 4, wherein the sorting weak knowledge points of the target user to obtain the weak knowledge point sorting of the target user comprises:

comparing the weak knowledge points of the target user with the weak knowledge points of the target user, determining knowledge points mastered by the target user in the weak knowledge points of the target user as weak knowledge points which are preferentially solved by the target user, and determining knowledge points not mastered by the target user in the weak knowledge points of the target user as weak knowledge points which are solved by the target user after delay;

and obtaining weak knowledge point sequencing of the target user according to the weak knowledge points which are solved by the target user preferentially and the weak knowledge points which are solved later.

7. The method of claim 6, wherein the obtaining the rank of weak knowledge points of the target user according to weak knowledge points preferentially solved and weak knowledge points later solved by the target user comprises:

obtaining initial sequencing of weak knowledge points of the target user according to the weak knowledge points which are solved by the target user preferentially and the weak knowledge points which are solved later;

acquiring second learning situation data of a target area where the target user is located, and acquiring weak knowledge points of the target area and weights of the weak knowledge points based on the second learning situation data;

and adjusting the initial sequencing of the weak knowledge points of the target user according to the weak knowledge points of the target area and the weights of the weak knowledge points to obtain the sequencing of the weak knowledge points of the target user.

8. The method according to any one of claims 1 to 4, wherein the determining the learning path of the target user for the weak knowledge points according to the weak knowledge point ranking comprises:

determining characteristic information output by a middle layer of the vulnerability diagnosis model as a learning-style picture of the target user;

inputting the learning situation picture of the target user, the weak knowledge point sequence and the learning resources of the target area where the target user is located into a path representation model to obtain the learning path representation of the target user;

and obtaining the learning path of the target user to the weak knowledge points according to the learning path representation of the target user and the preset learning path representations of N users, wherein N is a positive integer.

9. The method according to claim 8, wherein the obtaining of the learning path of the target user for the weak knowledge points according to the learning path characterization of the target user and the learning path characterizations of the preset N users comprises:

determining a similarity between the learned path characterization of the target user and the learned path characterization of each of the N users;

obtaining M users with the similarity meeting a preset condition from the N users, wherein M is a positive integer less than or equal to N;

and determining the learning path of one user in the M users as the learning path of the target user to the weak knowledge point.

10. The method of claim 9, wherein the determining the learning path of one of the M users as the learning path of the target user for weak knowledge points comprises:

determining a target learning path representation with highest similarity and minimum cost with the learning path representations of the target users in the learning path representations of the M users;

and determining the learning path of the target user to the weak knowledge points according to the learning path corresponding to the target learning path characterization.

11. The method according to claim 10, wherein the determining the learning path of the target user for the weak knowledge points according to the corresponding learning path characterization of the target learning path comprises:

the target learning path represents a corresponding learning path and is determined as an initial learning path of the target user to weak knowledge points;

acquiring a real-time answering result of the target user, taking the real-time answering result as an incentive, and updating the initial learning path through a reinforcement learning model;

and determining the updated initial learning path as the learning path of the target user to the weak knowledge points.

12. The method according to claim 8, wherein the inputting the learning emotion image of the target user, the weak knowledge point sequence and the learning resource of the target area where the target user is located into a path characterization model to obtain the learning path characterization of the target user comprises:

acquiring a real-time answering result of the target user, and mapping the real-time answering result into an answering result vector through a language model;

and inputting the learning situation picture, the weak knowledge point sequence, the learning resource and the answering result vector into the path representation model to obtain the learning path representation of the target user.

13. The method according to any one of claims 1-4, further comprising:

acquiring historical situation data of K students, wherein K is a positive integer;

inputting the historical situation data of the K students into the weak point diagnosis model to obtain a predicted weak knowledge point of each student of the K students;

determining a loss between the predicted weak knowledge points of each of the K students and the true weak knowledge points of each of the K students;

and training the weak point diagnosis model according to the loss.

14. The method according to any one of claims 1 to 4, wherein the vulnerability diagnostic model is a time series model.

15. A learning path determination device characterized by comprising:

16. A computing device, comprising: a memory, a processor;

the memory for storing a computer program;

the processor for executing the computer program to implement the method of any one of the preceding claims 1 to 14.

17. A computer-readable storage medium having computer-executable instructions stored thereon, which when executed by a processor, are configured to implement the method of any one of claims 1 to 14.