A sentiment-enhanced knowledge tracking method and system based on a graph neural network and a storage medium

Abstract

The application discloses a kind of based on graph neural network sentiment enhancement knowledge tracking method, system and storage medium, method includes according to the problem embedding of learner's exercise interaction information, skill embedding and answer embedding, and further constructs problem-skill heterostructure graph, according to the sentiment embedding of learner's sentiment state information;By the feature extraction of problem-skill heterostructure graph to question-skill embedding, which is obtained by trained graph convolution network model;According to question-skill embedding, answer embedding and sentiment embedding, obtain final exercise embedding;The final exercise embedding and sentiment embedding of each time step are input into the trained knowledge state evolution model, and the current knowledge state of learner is output;According to the answer result of predicted problem of learner's current knowledge state prediction.This application constructs knowledge tracking model by the way of sentiment embedding enhancement, improves prediction accuracy, and can be widely applied in intelligent education technology field.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent education, and in particular to a graph neural network-based emotion-enhanced knowledge tracking method and system and a storage medium. BACKGROUND

[0002] Knowledge tracking is a method for modeling the changes in students' knowledge states and is the basis for realizing personalized learning. Most current mainstream graph neural network-based emotion-enhanced knowledge tracking methods do not consider the influence of learners' emotional factors on the results. However, emotional factors can greatly affect learners' learning performance, and ignoring the influence of emotional factors can lead to the modeling of non-real levels of learners by the knowledge tracking task, reducing the effectiveness of the model. SUMMARY

[0003] The present application aims to at least partially solve one of the problems in the prior art.

[0004] To this end, an object of an embodiment of the present application is to provide a graph neural network-based emotion-enhanced knowledge tracking method that jointly models learners' knowledge states and emotional states by combining a graph convolution network model (GCN) and a Transformer model, taking into account learners' emotional state information and time series information of knowledge evolution, and effectively improving the prediction accuracy of the knowledge tracking model.

[0005] Another object of an embodiment of the present application is to provide a graph neural network-based emotion-enhanced knowledge tracking method system.

[0006] To achieve the above technical objects, the technical solutions adopted by the embodiments of the present application include:

[0007] In a first aspect, the embodiments of the present application provide a graph neural network-based emotion-enhanced knowledge tracking method, characterized by comprising the following steps:

[0008] Obtaining learners' problem interaction information and emotional state information;

[0009] According to the problem interaction information, obtaining problem embeddings, skill embeddings, and answer embeddings, and according to the emotional state information, obtaining emotional embeddings;

[0010] Constructing a problem-skill heterogeneous graph according to the problem embeddings and the skill embeddings, and extracting features from the problem-skill heterogeneous graph by a trained graph convolution network model to obtain problem-skill embeddings;

[0011] Obtaining a final practice embedding according to the problem-skill embeddings, the answer embeddings, and the emotional embeddings;

[0012] inputting the final exercise embedding and the emotion embedding of each time step into a trained knowledge state evolution model, and outputting a current knowledge state of the learner;

[0013] generating an answer result of a to-be-predicted question according to the current knowledge state of the learner and the to-be-predicted question.

[0014] Further, the exercise interaction information includes a student set, a question set, a skill set, an answer set, and an answer time stamp information of the learner.

[0015] Further, the obtaining of a question embedding, a skill embedding, and an answer embedding according to the exercise interaction information, and the obtaining of an emotion embedding according to the emotion state information include:

[0016] embedding processing is performed on the question set, the skill set, and the answer set to obtain the question embedding, the skill embedding, and the answer embedding;

[0017] discretization and standardization processing is performed on the emotion state information to obtain the emotion embedding.

[0018] Further, the constructing of a question-skill heterogeneous graph according to the question embedding and the skill embedding includes:

[0019] constructing a question-skill bipartite graph according to the question set and the skill set;

[0020] analyzing the relationship between the question set and the skill set by an association rule mining algorithm according to the question-skill bipartite graph to construct the question-skill heterogeneous graph.

[0021] Further, the obtaining of a final exercise embedding according to the question-skill embedding, the answer embedding, and the emotion embedding includes:

[0022] aggregating the question-skill embedding, the answer embedding, and the emotion embedding to obtain an aggregated embedding;

[0023] performing nonlinear transformation on the aggregated embedding to obtain the final exercise embedding.

[0024] Further, the generating of an answer result of a to-be-predicted question according to the current knowledge state of the learner and the to-be-predicted question includes:

[0025] processing the to-be-predicted question by the trained graph convolution network model to obtain a to-be-predicted question embedding;

[0026] The current knowledge state and the embedding of the to-be-predicted question are combined and nonlinearly transformed through a fully connected network and an activation function to obtain a response result of the to-be-predicted question.

[0027] Further, the graph convolution network model and the knowledge state evolution model are trained through the following steps:

[0028] The predicted response result of the to-be-predicted question is compared with an actual response result through a loss function to obtain a loss value.

[0029] According to the loss value, parameters of the graph convolution network model and the knowledge state evolution model are adjusted through a back propagation algorithm.

[0030] In a second aspect, an embodiment of the present application provides a sentiment-enhanced knowledge tracking system based on a graph neural network, and the system comprises:

[0031] A data acquisition module is configured to acquire exercise interaction information and sentiment state information of a learner.

[0032] A data preprocessing module is configured to obtain question embedding, skill embedding, and response embedding according to the exercise interaction information, and obtain sentiment embedding according to the sentiment state information.

[0033] A feature extraction module is configured to construct a question-skill heterogeneous graph according to the question embedding and the skill embedding, and obtain question-skill embedding by performing feature extraction on the question-skill heterogeneous graph through a trained graph convolution network model.

[0034] A feature fusion module is configured to obtain final exercise embedding according to the question-skill embedding, the response embedding, and the sentiment embedding.

[0035] A knowledge state prediction module is configured to input the final exercise embedding and the sentiment embedding of each time step into a trained knowledge state evolution model to output a current knowledge state of the learner.

[0036] A response result prediction module is configured to generate a response result of a to-be-predicted question according to the current knowledge state of the learner and the to-be-predicted question.

[0037] In a third aspect, an embodiment of the present application provides a device, and the device comprises:

[0038] At least one processor;

[0039] At least one memory configured to store at least one program;

[0040] When the at least one program is executed by the at least one processor, the at least one processor implements the emotion-enhanced knowledge tracking method based on the graph neural network as described above.

[0041] In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium, wherein a processor executable program is stored, and the processor executable program is used for executing the emotion-enhanced knowledge tracking method based on the graph neural network when executed by a processor.

[0042] The advantages and beneficial effects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be understood through the practice of the present application:

[0043] The present application constructs a question-skill heterogeneous graph by analyzing the mutual relationship between the question set and the skill set, uses a graph convolutional neural network to extract high-order features of the question-skill heterogeneous graph, obtains the knowledge state features of the learner considering the emotional state information in an all-round way after fusing the answer embedding and the emotional embedding, and takes the knowledge state features and the emotional state features at each time step as the input of a Transformer model to complete the modeling of the knowledge evolution model based on the self-attention mechanism, which can comprehensively and dynamically evaluate the knowledge state and emotional fluctuation of the learner, effectively predict the answering situation of the learner, and improve the prediction accuracy of the knowledge tracking model. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 A step diagram of the emotion-enhanced knowledge tracking method based on the graph neural network provided by the embodiment of the present application is provided.

[0045] Figure 2 A technical scheme flow diagram of the emotion-enhanced knowledge tracking method based on the graph neural network provided by the embodiment of the present application is provided.

[0046] Figure 3 A system diagram of the emotion-enhanced knowledge tracking system based on the graph neural network provided by the embodiment of the present application is provided.

[0047] Figure 4 A structure diagram of the computer device provided by the embodiment of the present application is provided. DETAILED DESCRIPTION

[0048] Embodiments of the present application are described below in detail, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary, only for the purpose of explaining the present application, and cannot be understood as limiting the present application. For the step numbers in the following embodiments, they are only set for the convenience of setting out the description, and the order between the steps is not limited in any way, and the execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0049] In the description of the present application, the meaning of multiple is two or more, and if the first, the second is described, it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features or implicitly indicating the sequence of indicated technical features. In addition, unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art.

[0050] Figure 1 A step diagram of a sentiment-enhanced knowledge tracking method based on a graph neural network provided by an embodiment of the present application is provided with reference to Figure 1 The embodiment of the present application provides a sentiment-enhanced knowledge tracking method based on a graph neural network, characterized in that the method comprises the following steps:

[0051] Obtain the exercise interaction information and the sentiment state information of the learner;

[0052] According to the exercise interaction information, obtain the question embedding, the skill embedding and the answer embedding, and according to the sentiment state information, obtain the sentiment embedding;

[0053] According to the question embedding and the skill embedding, construct a question-skill heterogeneous graph, and through the trained graph convolution network model, extract features of the question-skill heterogeneous graph to obtain question-skill embedding;

[0054] According to the question-skill embedding, the answer embedding and the sentiment embedding, obtain the final exercise embedding;

[0055] Input the final exercise embedding and the sentiment embedding of each time step into the trained knowledge state evolution model, and output the current knowledge state of the learner;

[0056] According to the current knowledge state of the learner and the problem to be predicted, generate the answer result of the problem to be predicted.

[0057] Specifically, specifically, Figure 2 A technical solution flowchart of a sentiment-enhanced knowledge tracking method based on a graph neural network provided by the embodiment is provided, and the following will be described in combination with Figure 2The implementation process of the technical solution of the embodiment is introduced. In the embodiment, by analyzing the practice data of the students, the student set, the question set, the skill set, the answer set and the answer time stamp information and other information can be obtained, and the emotional state information of the students can be obtained by asking according to the questionnaire and combining the computer vision during answering the questions. The time stamp information of the emotional state and the time stamp information of the answering are aligned, and the emotional information corresponding to each answering data can be obtained.

[0058] In the embodiment, after obtaining the exercise interaction information and the emotional state information of the learners, the embedding layer is processed to obtain the embedding representation of the student questions, skills, answers and emotions in the set (q i ,c i ,a i ,e i ). There are four embedding matrices in total:

[0059]

[0060] Among them, |Q| represents the size of the question set, that is, the total number of all questions, |S| represents the size of the skill set, that is, the total number of all skills, |E| represents the number of emotional states, and the second dimension d represents the feature dimension of the embedding. The data generated by the learners during the learning process is processed by embedding to obtain the corresponding embedding representation (q i ,c i ,a i ,e i ).

[0061] After embedding processing, the student set question set skill set and emotional state set are obtained. N s , N q , N c and E s represent the number of students, questions, skills and emotions respectively. It is assumed that at time t, the knowledge state of the student is h t , h t represents the mastery degree of the student to each knowledge.

[0062] In the embodiment, after obtaining the embedding representation of each index, the knowledge correlation matrix QC is constructed according to the known prior relationship between the questions and the skills. QC is a matrix containing only 0 and 1. When the question q i contains the knowledge concept c k , QC jk =1, otherwise QC jk= 0, j and k represent the row number and column number corresponding to the QC respectively. By aligning the timestamp information of the emotional state and the timestamp information of the answer, the emotional state e of the student corresponding to each answer data can be obtained i . Thus, the cognitive data of the student in the learning process can be represented as a historical cognitive record set:

[0063] {(q1, c1, a1), e1, (q2, c2, a2), e2, …, (q i , i , i , i , t , t , t , t}

[0064] wherein a t ∈ {0, 1} represents whether the student correctly answers the question q t at time t.

[0065] In the embodiment, from the perspective of reasoning, whether the student can correctly answer the new question depends on the mastery of the relevant skills and the characteristics of the question. If the student has solved similar questions before, he is more likely to correctly answer the new question. Since the problem-skill relationship graph is bidirectional, the first-hop neighbor of a question should be its corresponding skill, and the second-hop neighbor should be other questions sharing the same skill, therefore, the knowledge correlation matrix QC can only capture the direct correlation between the question and the skill, and cannot capture the deeper implicit information between the question and the question, and between the skill and the skill. Therefore, in the embodiment, according to the constructed knowledge correlation matrix QC, the relationship between the question and the skill (knowledge concept) is modeled as a heterogeneous graph, and the node features are aggregated layer by layer through a graph convolution network (GCN) to obtain more rich knowledge representation, and the problem embedding and the skill embedding related to the skill and the question are updated through embedding propagation by using the graph convolution network, thereby extracting high-order feature information,

[0066] The graph convolution network stacks multiple graph convolution layers to encode high-order adjacent information, and in each layer, the node representation can be updated through the embedding of itself and adjacent nodes. The representation of node i in the graph is denoted as x i (x i can represent skill embedding c i or question embedding q i , and the neighbor node set is denoted as N i ), and the formula of the l-th layer GCN is:

[0067]

[0068] wherein, It is the representation of node i in the l-th layer, N i W is the set of neighbors of node i. (l) and b (l) These are the weight matrix and bias of the l-th layer, respectively, and σ represents the activation function, which can be the ReLU function in this embodiment.

[0069] In this embodiment, after extracting the high-order knowledge state features through the above processing, it is necessary to subsequently combine these knowledge state features with the sentiment embedding vector e. t Answer embedded in a t And the problem embedding obtained by graph convolutional networks To enhance the model's ability to perceive changes in students' emotions, a feature fusion module is used to fuse questions, answers, and emotions, resulting in an aggregated embedding. This aggregated embedding is then passed through a non-linear neural network to obtain the final practice embedding. The calculation formula for the entire process is as follows:

[0070]

[0071] Where, x t This is the output of the fusion module, representing the student's comprehensive vector considering emotional factors. The question embedding is obtained through a Graph Convolutional Network (GCN). t It is an answer to embedding, e t It is an emotion embedding, where W1 and b1 are weight matrices and bias vectors used to integrate different embeddings and map them to the same dimension.

[0072] In this embodiment, after obtaining the final practice embedding information, a knowledge state evolution model based on the self-attention mechanism will be constructed based on the Transformer model.

[0073] To overcome the limitations of traditional LSTM in long sequence modeling, the Transformer model can be used to evolve students' knowledge states, thereby overcoming the limitations of traditional LSTM in long sequence modeling and modeling long-term dependencies in students' learning sequences.

[0074] In the self-attention module, the input interaction sequence X is first transformed linearly to obtain the query matrix Q, key matrix K, and value matrix V, and then the attention weights A are calculated:

[0075]

[0076] Where, d k The dimension of the key vector is used to scale the attention score. Through parallel computation of multiple attention heads, the Transformer can effectively capture long-term dependencies and complex relationships.

[0077] In combination with the emotional state information, by embedding the emotional state of the learner at each time step into the input of the Transformer, a knowledge state representation enhanced by emotion can be formed:

[0078] h t =Transformer([x,x2,…,x t ,e t ])

[0079] Where e t represents the emotional state embedding of the learner at time step t, through the fusion of emotional state, the understanding of the current knowledge state of the learner is enhanced.

[0080] In this embodiment, after obtaining the current knowledge state h t of the learner, the interaction between the current state and the problem to be predicted is used to generate the final answer prediction result. Specifically, a fully connected network is used and activated by a sigmoid function to predict the correct probability of the learner's answer to the new question:

[0081]

[0082] Where W out is the weight matrix of the output layer, and b out is the bias term.

[0083] In this embodiment, after completing the joint modeling based on the graph convolution network model and the Transformer model, the historical learning data and emotional state of the learner are used to train the model using the cross-entropy loss function:

[0084]

[0085] Where y i is the actual label, is the prediction probability of the model. By optimizing the loss function, the parameters in the graph neural network and the Transformer module are updated to improve the prediction accuracy and generalization ability.

[0086] ​It can be recognized that the embodiment constructs a question-skill heterogeneous graph by analyzing the correlation between the question set and the skill set, extracts high-order features of the question-skill heterogeneous graph using a graph convolutional neural network, obtains a knowledge state feature of a learner considering the emotional state information in a comprehensive manner after fusing the answer embedding and the emotional embedding, and completes the modeling of the knowledge evolution model based on the self-attention mechanism by taking the knowledge state feature and the emotional state feature at each time step as the input of the Transformer model, thereby being capable of comprehensively and dynamically evaluating the knowledge state and emotional fluctuation of the learner, effectively predicting the answering situation of the learner, and improving the prediction accuracy of the knowledge tracking model.

[0087] In some alternative embodiments, the exercise interaction information includes a student set of learners, a question set, a skill set, an answer set, and an answering timestamp information.

[0088] Specifically, in the present embodiment, the student set is mainly used to distinguish different learners, so as to ensure that the model can track knowledge based on the historical learning records of different learners; the question set and the skill set include all questions and corresponding knowledge points, and the correspondence between the two can be obtained through pre-annotation, which is used for subsequent construction of the question-skill heterogeneous graph; the answer set is used to obtain the answering situation of the learner, and the deviation between the predicted answer result and the real answer result is used to calculate the loss value, and the parameters of the model are updated through the back propagation algorithm; and the answering timestamp information is used to align the emotional state and the answer result of the student, thereby providing data support for subsequent time series modeling.

[0089] In some alternative embodiments, according to the exercise interaction information, the question embedding, the skill embedding, and the answer embedding are obtained, and the emotional embedding is obtained according to the emotional state information, which includes:

[0090] The question set, the skill set, and the answer set are embedded to obtain the question embedding, the skill embedding, and the answer embedding;

[0091] The emotional state information is discretized and standardized to obtain the emotional embedding.

[0092] Specifically, in the present embodiment, the embedding processing mode is mainly to map the discrete high-dimensional data in each set to a low-dimensional vector space, such as mapping the discrete question number, skill number, and answer result data to corresponding vectors through an embedding matrix. As for the emotional state information, the emotional state categories can be divided into several possible types in advance, such as anxiety, calmness, etc. The division is based on the fact that there is a clear substantive difference between the categories, which is easy to distinguish, and the emotional state of different types may have a relatively obvious influence on the answering situation. After dividing the categories, the emotional state is mapped to discrete numerical data, and then standardized to ensure the uniformity of the data range.

[0093] In some alternative embodiments, constructing the question-skill heterogeneous graph according to the question embedding and the skill embedding comprises:

[0094] constructing a question-skill bipartite graph according to the question set and the skill set;

[0095] constructing a question-skill heterogeneous graph by a correlation rule mining algorithm to analyze the relationship between the question set and the skill set according to the question-skill bipartite graph.

[0096] In some alternative embodiments, obtaining the final practice embedding according to the question-skill embedding, the answer embedding and the sentiment embedding comprises:

[0097] aggregating the question-skill embedding, the answer embedding and the sentiment embedding to obtain an aggregated embedding;

[0098] performing a nonlinear transformation on the aggregated embedding to obtain the final practice embedding.

[0099] In some alternative embodiments, generating an answer result of a to-be-predicted question according to the current knowledge state of the learner and the to-be-predicted question comprises:

[0100] processing the to-be-predicted question by the trained graph convolution network model to obtain a to-be-predicted question embedding;

[0101] combining and performing a nonlinear transformation on the current knowledge state and the to-be-predicted question embedding by a fully connected network and an activation function to obtain the answer result of the to-be-predicted question.

[0102] In some alternative embodiments, the graph convolution network model and the knowledge state evolution model are trained by the following steps:

[0103] comparing the answer result prediction result of the to-be-predicted question with the actual answer result by a loss function to obtain a loss value;

[0104] adjusting the parameters of the graph convolution network model and the knowledge state evolution model by a back propagation algorithm according to the loss value.

[0105] Referring to Figure 3 The embodiment of the present application provides a sentiment-enhanced knowledge tracking system based on a graph neural network, which is characterized by comprising:

[0106] a data acquisition module configured to acquire problem interaction information and sentiment state information of a learner;

[0107] a data preprocessing module configured to obtain question embedding, skill embedding and answer embedding according to the problem interaction information, and obtain sentiment embedding according to the sentiment state information;

[0108] The feature extraction module is configured to construct a question-skill heterogeneous graph according to the question embedding and the skill embedding, and perform feature extraction on the question-skill heterogeneous graph by using a trained graph convolution network model to obtain question-skill embedding;

[0109] The feature fusion module is configured to obtain final practice embedding according to the question-skill embedding, the answer embedding, and the sentiment embedding.

[0110] The knowledge state prediction module is configured to input the final practice embedding and the sentiment embedding of each time step into a trained knowledge state evolution model to output a current knowledge state of the learner.

[0111] The answer result prediction module is configured to generate an answer result of the to-be-predicted question according to the current knowledge state of the learner and the to-be-predicted question.

[0112] Reference Figure 4 The embodiment of the present application provides a computer device, which comprises:

[0113] At least one processor;

[0114] At least one memory for storing at least one program;

[0115] When the above at least one program is executed by the above at least one processor, the above at least one processor implements the above method.

[0116] The content in the above method embodiment is applicable to the device embodiment, the device embodiment specifically implements the same functions as the above method embodiment, and achieves the same beneficial effects as the above method embodiment.

[0117] The embodiment of the present application further provides a computer readable storage medium, which stores a program executable by a processor, and the program executable by the processor is used to execute the above method when executed by the processor.

[0118] The computer readable storage medium of the embodiment of the present application can execute the method provided by the embodiment of the present application, and can execute any combination of the method embodiments to implement the steps, and has the corresponding functions and beneficial effects of the method.

[0119] The embodiment of the present application further discloses a computer program product or a computer program, which comprises computer instructions stored in a computer readable storage medium. The processor of the computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute Figure 1The emotion-enhanced knowledge tracing method based on a graph neural network is shown.

[0120] In some alternative embodiments, the functions / operations referred to in the block diagrams can not occur in the order shown in the operational illustrations. For example, two blocks shown in succession can in fact be executed substantially concurrently or the blocks can sometimes be executed in the reverse order, depending upon the functionality / operations involved. Also, the embodiments presented and described in the flowcharts are only examples. Alternative embodiments can be implemented where various operations are changed, omitted, or added. Accordingly, the disclosed methods are not limited to the order of operations presented in the flowcharts.

[0121] Furthermore, although the present application is described in the context of functional modules, it is understood that one or more of the functions and / or features described above can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules, unless otherwise specified. It is also understood that detailed discussion of the actual implementation of each module is unnecessary to an understanding of the present application. Rather, the actual implementation is within the routine skill of engineers familiar with the attributes, functions, and internal relationships of the various functional modules disclosed in the devices disclosed herein. Accordingly, the present application is not limited to the details of the implementations described herein, but rather is limited only by the scope of the claims and the equivalents thereof. It is also understood that the particular concepts disclosed are illustrative only and not intended to limit the scope of the present application, which is defined only by the claims and equivalents thereof.

[0122] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the part of the technical solutions that make essential contributions to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the above-mentioned methods according to the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0123] The logic and / or steps represented in the flow diagrams or otherwise described herein, for example, can be embodied in non-transitory computer- readable media, executed by an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or in conjunction with which the instructions can be executed. In the context of this specification, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium.

[0124] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can also be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and stored in a computer memory.

[0125] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or combinations thereof, can be used: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), and / or the like.

[0126] In the above description of the present specification, reference is made to the description of terms such as "one embodiment / one example", "another embodiment / another example", or "certain embodiments / certain examples" and the like, which means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. The illustrative appearances of the above-mentioned terms in the description are not necessarily referred to the same embodiment or example throughout the specification. Moreover, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0127] While the embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary and are not to be construed as limiting the scope of the application. The scope of the application is defined by the appended claims and their equivalents.

[0128] The above is the specific description of the preferred embodiment of the application, but the application is not limited to the above-mentioned embodiments, and those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the application.

Claims

1. A graph neural network-based emotion-enhanced knowledge tracking method, characterized in that, The method comprises the following steps: Obtaining exercise interaction information and emotional state information of a learner; According to the exercise interaction information, obtaining question embedding, skill embedding and answer embedding, and according to the emotional state information, obtaining emotional embedding; According to the question embedding and the skill embedding, constructing a question-skill heterogeneous graph, and performing feature extraction on the question-skill heterogeneous graph through a trained graph convolution network model to obtain question-skill embedding; According to the question-skill embedding, the answer embedding and the emotional embedding, obtaining a final practice embedding; Inputting the final practice embedding and the emotional embedding of each time step into a trained knowledge state evolution model to output the current knowledge state of the learner; According to the current knowledge state of the learner and a to-be-predicted question, generating an answer result of the to-be-predicted question; The exercise interaction information comprises a student set, a question set, a skill set, an answer set and a timestamp information of the learner; The question-skill heterogeneous graph is constructed according to the question set and the skill set, and the relationship between the question set and the skill set is analyzed through an association rule mining algorithm to construct the question-skill heterogeneous graph; The final practice embedding is obtained according to the question-skill embedding, the answer embedding and the emotional embedding, which comprises: Aggregating the question-skill embedding, the answer embedding and the emotional embedding to obtain an aggregated embedding; Performing nonlinear transformation on the aggregated embedding to obtain the final practice embedding. The question embedding, the skill embedding and the answer embedding are obtained according to the exercise interaction information, and the emotional embedding is obtained according to the emotional state information, which comprises: Performing embedding processing on the question set, the skill set and the answer set to obtain the question embedding, the skill embedding and the answer embedding; 2. The emotion-enhanced knowledge tracing method based on a graph neural network according to claim 1, characterized in that, Discretizing and standardizing the emotional state information to obtain the emotional embedding. The answer result of the to-be-predicted question is generated according to the current knowledge state of the learner and the to-be-predicted question, which comprises: Processing the to-be-predicted question through the trained graph convolution network model to obtain a to-be-predicted question embedding; 3. The emotion-enhanced knowledge tracing method based on a graph neural network according to claim 1, characterized in that, Combining and nonlinearly transforming the current knowledge state and the to-be-predicted question embedding through a full connection network and an activation function to obtain the answer result of the to-be-predicted question. The graph convolution network model and the knowledge state evolution model are trained through the following steps: Comparing the answer result prediction result of the to-be-predicted question with the actual answer result through a loss function to obtain a loss value; 4. The emotion-enhanced knowledge tracing method based on a graph neural network according to claim 1, characterized in that, According to the loss value, adjusting the parameters of the graph convolution network model and the knowledge state evolution model through a back propagation algorithm. The method comprises the following steps: A data acquisition module is configured to obtain exercise interaction information and emotional state information of a learner; 5.A graph neural network based sentiment-enhanced knowledge tracing system, characterized in that, A data preprocessing module is configured to obtain question embedding, skill embedding and answer embedding according to the exercise interaction information, and to obtain emotional embedding according to the emotional state information; ​ ​ The feature extraction module is configured to construct a question-skill heterogeneous graph according to the question embedding and the skill embedding, and perform feature extraction on the question-skill heterogeneous graph by using a trained graph convolution network model to obtain a question-skill embedding; The feature fusion module is configured to obtain a final practice embedding according to the question-skill embedding, the answer embedding, and the sentiment embedding; The knowledge state prediction module is configured to input the final practice embedding and the sentiment embedding of each time step into a trained knowledge state evolution model to output a current knowledge state of the learner; The answer result prediction module is configured to generate an answer result of a to-be-predicted question according to the current knowledge state of the learner and the to-be-predicted question. The exercise interaction information includes a student set, a question set, a skill set, an answer set, and answer time stamp information of the learner. The question-skill heterogeneous graph is constructed according to the question embedding and the skill embedding, including: A question-skill bipartite graph is constructed according to the question set and the skill set; According to the question-skill bipartite graph, a relationship between the question set and the skill set is analyzed by using an association rule mining algorithm to construct the question-skill heterogeneous graph; The final practice embedding is obtained according to the question-skill embedding, the answer embedding, and the sentiment embedding, including: An aggregated embedding is obtained by aggregating the question-skill embedding, the answer embedding, and the sentiment embedding; The aggregated embedding is subjected to nonlinear transformation to obtain the final practice embedding.

6. An electronic device, comprising: The apparatus includes: at least one processor; at least one memory configured to store at least one program; when the at least one program is executed by the at least one processor, the at least one processor is caused to implement the emotion-enhanced knowledge tracking method based on a graph neural network according to any one of claims 1-4.

7. A computer readable storage medium having stored therein a program that is executable by a processor, characterized in that, The processor-executable program, when executed by the processor, is configured to perform the emotion-enhanced knowledge tracking method based on a graph neural network according to any one of claims 1-4.

Images