Migration knowledge tracking method, system and storage medium in sparse data scenarios

Abstract

The present invention relates to the field of intelligent education technology, and discloses a migration knowledge tracking method, system and storage medium in a sparse data scenario. The construction process of the knowledge tracking model adopted by the method includes: selecting the optimal dense data space by calculating the correlation between the sparse data space and different dense data spaces; performing sequence difference analysis on question embedding and interaction embedding to obtain guiding knowledge variables; modeling the student's interaction records to generate the student's knowledge state variables; after fusing the guiding knowledge variables and the knowledge state variables, they are input into the classifier together with the question embedding at the next moment to predict the probability of the student answering the question correctly at the next moment; freezing the parameters of the knowledge state navigation module of the optimal dense data space, and migrating it to the sparse data space, so as to perform cross-training. In the sparse data scenario, the model can use the migrated knowledge to make up for the lack of local data and significantly improve the prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the field of intelligent education technology, and in particular to a migration knowledge tracking method, system and storage medium in a sparse data scenario. Background Art

[0002] Knowledge tracing is a method used to assess and predict students' understanding level, mainly by analyzing their performance in exercises. The basic goal of knowledge tracing is to provide more accurate assessments for individual learners, and then provide tailored personalized educational materials, which is of great significance to the development of adaptive learning environments. In the field of sequence-based knowledge tracing, the Attentive Knowledge Tracing (AKT) method is regarded as a key technology. This method is popular due to its three advantages: fewer parameters, fast processing speed, and superior performance. Recently, many cutting-edge methods have emerged, such as using contrastive learning techniques to reveal semantic similarities or differences in learning history to deepen the understanding of the relationship between students and questions; or using two innovative attention-based models to clarify the associations between elements in the input sequence, and solving the prediction instability problem in deep knowledge tracing (DKT) through finite state automata (FSA).

[0003] Transfer learning applies the knowledge learned from one task to another task, while domain adaptation (DA) aims to solve the difference between the source domain and the target domain. It adapts the source domain model with rich data to the target domain with scarce data through parameter transfer.

[0004] Early domain adaptation methods used statistics such as maximum mean difference (MMD) and correlation alignment (CORAL) to align feature distributions. Deep learning technology has enhanced domain adaptation capabilities through domain adversarial neural networks and adaptive batch normalization, and uses adversarial training to adjust deep network layers to adapt to the statistical characteristics of the source and target domains.

[0005] Recently, contrastive learning and self-supervised learning have been introduced into domain adaptation, enabling models to quickly adapt to new domains under unsupervised training, and accelerating this process with the help of pre-training knowledge to ensure that the model has good generalization capabilities in the target domain.

[0006] The technical problem to be solved by the present invention is that in sparse data scenarios, knowledge tracking models face challenges such as scarce data and short interaction sequences, resulting in low prediction accuracy of knowledge tracking models. Most existing knowledge tracking methods rely on abundant data to train models, but in sparse data environments, these knowledge tracking methods are difficult to apply effectively. Summary of the invention

[0007] To solve the above technical problems, the present invention provides a migration knowledge tracking method, system and storage medium in a sparse data scenario, which migrates knowledge in a dense data space to a sparse data space by means of domain migration to improve the performance of the knowledge tracking model in the sparse data space, thereby solving the challenges of knowledge tracking in sparse data scenarios.

[0008] In order to solve the above technical problems, the present invention adopts the following technical solutions: A migration knowledge tracking method in a sparse data scenario, the construction process of the adopted knowledge tracking model includes: Encode the topics and interaction records in the sparse data set and multiple dense data sets into a sparse data space and multiple dense data spaces respectively, and both the sparse data space and the dense data space contain topic embedding and interaction embedding; The correlation between the sparse data space and different dense data spaces is calculated through the weighted average difference of the question embedding, and the optimal dense data space is selected for the sparse data space; The knowledge tracking task is carried out in the optimal dense data space through the knowledge state navigation module and the knowledge sequence analysis module; the knowledge state navigation module of the optimal dense data space is frozen, and the sequence difference analysis of the question embedding and the interaction embedding is performed to obtain the guiding knowledge variables; the knowledge sequence analysis module of the sparse data space is used to model the students' interaction records and generate the students' knowledge state variables; After the guiding knowledge variable and the knowledge state variable are fused, they are input into the classifier together with the embedding of the question at the next moment to predict the probability that the student will answer the question correctly at the next moment. Cross-training is performed by freezing the parameters of the knowledge state navigation module in the optimal dense data space and migrating it to the sparse data space.

[0009] In one embodiment, the encoding of topics and interaction records in a sparse data set and a plurality of dense data sets into a sparse data space and a plurality of dense data spaces, respectively, wherein both the sparse data space and the dense data space contain topic embeddings and interaction embeddings, specifically includes: Through the selected knowledge tracking model, the questions and interaction records in the sparse data set and dense data set are encoded to obtain sparse data space and dense data space; in the sparse data space and dense data space, each question and interaction record is represented as question embedding and interaction embedding respectively, which capture the characteristics of the question and the relevant information of the interaction with the students.

[0010] In one embodiment, calculating the correlation between the sparse data space and different dense data spaces by weighted average difference of the topic embeddings and selecting the optimal dense data space for the sparse data space specifically includes: All the topic embeddings in the sparse data space or dense data space are subtracted pairwise to obtain difference embeddings, and the central characteristics of the sparse data space or dense data space are obtained by weighted average difference embedding. The correlation between the central characteristics of the sparse data space and the central characteristics of different dense data spaces is compared by cosine similarity, and the dense data space corresponding to the highest correlation is selected as the optimal dense data space of the sparse data space.

[0011] In one embodiment, the sequence difference analysis of the topic embedding and the interaction embedding is performed to obtain the guiding knowledge variable, which specifically includes: In the optimal dense data space, the offset subtraction operation is performed on the question embedding sequence and the interaction embedding sequence to generate the question difference sequence and the interaction difference sequence; Input the question difference sequence into a first decoder and output question comprehension features; Input the interaction difference sequence into the second decoder and output the learning behavior feature; The question comprehension features are used as queries and keys, and the learning behavior features are used as values, which are input into the third decoder to generate guiding knowledge variables.

[0012] In one embodiment, the first decoder, the second decoder and the third decoder all adopt a monotonic multi-head attention mechanism.

[0013] In one embodiment, the knowledge sequence analysis module using sparse data space models the student's interaction records to generate the student's knowledge state variables, specifically including: Modeling student interaction record sequences in sparse data space ; They represent the questions answered by the students at time t and the variables used to indicate whether the answers given by the students are correct; the interaction record sequence is input into the pre-trained knowledge tracking model to generate the student knowledge state variable.

[0014] In one embodiment, the classifier adopts a multi-layer perceptron.

[0015] A migration knowledge tracking system in a sparse data scenario, comprising: Data space construction module: Encode the topics and interaction records in the sparse data set and multiple dense data sets into a sparse data space and multiple dense data spaces respectively. Both the sparse data space and the dense data space contain topic embedding and interaction embedding. Dense data space selection module: calculates the correlation between sparse data space and different dense data spaces through the weighted average difference of question embedding, and selects the optimal dense data space for the sparse data space; Knowledge tracking task module: The knowledge tracking task is performed in the optimal dense data space through the knowledge state navigation module and the knowledge sequence analysis module; the knowledge state navigation module in the optimal dense data space is frozen, and the sequence difference analysis of the question embedding and the interaction embedding is performed to obtain the guiding knowledge variables; the knowledge sequence analysis module in the sparse data space is used to model the students' interaction records and generate the students' knowledge state variables; Transfer prediction module: After fusing the guiding knowledge variables and knowledge state variables, they are input into the classifier together with the embedding of the question at the next moment to predict the probability that the student will answer the question correctly at the next moment; Model training module: Cross-training is performed by freezing the parameters of the knowledge state navigation module in the optimal dense data space and migrating it to the sparse data space.

[0016] In one embodiment, the dense data space selection module subtracts all the topic embeddings in the sparse data space or the dense data space from each other to obtain difference embeddings, obtains the central characteristics of the sparse data space or the dense data space by weighted average difference embeddings, uses cosine similarity to compare the correlation between the central characteristics of the sparse data space and the central characteristics of different dense data spaces, and selects the dense data space corresponding to the highest correlation as the optimal dense data space of the sparse data space.

[0017] The system of the present invention corresponds to the method, and the preferred embodiments applicable to the method are also applicable to the system.

[0018] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method in any one of the embodiments.

[0019] Compared with the prior art, the beneficial technical effects of the present invention are: The present invention extracts cross-domain common knowledge by screening the optimal dense data space, captures dynamic knowledge changes, and generates guiding knowledge variables. In sparse data scenarios, the model can use migrated knowledge to make up for the lack of local data and significantly improve prediction accuracy (such as the correctness of answering questions). BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of a knowledge tracking method in an embodiment of the present invention; Figure 2 is a schematic diagram of the architecture of a knowledge tracking model in an embodiment of the present invention; Figure 3 This is a schematic diagram showing that the central characteristic of the differential embedding for analyzing the topic in an embodiment of the present invention is more suitable for migration. DETAILED DESCRIPTION

[0021] A preferred embodiment of the present invention is described in detail below with reference to the accompanying drawings.

[0022] like Figure 1 As shown, the present invention provides a migration knowledge tracking method in a sparse data scenario, and the construction process of the adopted knowledge tracking model includes the following steps: S1, encode the topics and interaction records in the sparse data set and multiple dense data sets into a sparse data space and multiple dense data spaces respectively, and both the sparse data space and the dense data space contain topic embedding and interaction embedding; S2, calculates the correlation between the sparse data space and different dense data spaces through the weighted average difference of the question embedding, and selects the optimal dense data space for the sparse data space; S3, through the knowledge state navigation module and the knowledge sequence analysis module, the knowledge tracking task is carried out in the dense data space; the knowledge state navigation module of the optimal dense data space is frozen, and the sequence difference analysis of the question embedding and the interaction embedding is performed to obtain the guiding knowledge variables; the knowledge sequence analysis module of the sparse data space is used to model the students' interaction records and generate the students' knowledge state variables; S4, after fusing the guiding knowledge variable and the knowledge state variable, they are input into the classifier together with the question embedding of the next moment to predict the probability that the student will answer the question correctly at the next moment; S5, by freezing the parameters of the knowledge state navigation module in the optimal dense data space and migrating it to the sparse data space, cross-training is performed to improve the performance of the knowledge tracking model in the sparse data space.

[0023] The knowledge tracking model in the trained sparse data space can generate the probability that the student will correctly answer the question at the next moment based on the question sequence and interaction records in the student's actual learning process.

[0024] Dense data sets can use open source educational data sets (such as large-scale question banks and student answer records), but the amount of data on the answer records of a single student or a class or school is small, and the prediction accuracy of the knowledge tracking model trained with these sparse data is not high. The present invention uses a parameter freezing migration mechanism to migrate the parameters in the knowledge state guidance module in the dense data space to the sparse data space, and dynamically fuses the migrated knowledge variables (i.e., guiding knowledge variables) with the local knowledge state variables to adapt to the personalized characteristics of the target domain. This enables the knowledge tracking model to effectively handle unseen questions or knowledge points and reduce dependence on local data distribution.

[0025] In one embodiment, in step S1, the topics and interaction records in the sparse data set and the multiple dense data sets are encoded into a sparse data space and multiple dense data spaces respectively, and the sparse data space and the dense data space both contain topic embedding and interaction embedding, specifically including: Through the selected knowledge tracking model, the questions and interaction records in the sparse data set and dense data set are encoded to obtain sparse data space and dense data space; in the sparse data space and dense data space, each question and interaction record is represented as question embedding and interaction embedding respectively, which capture the characteristics of the question and the relevant information of the interaction with the students.

[0026] In this way, topics and interaction records are transformed from raw data form into embedded representations suitable for model processing, providing a basis for subsequent analysis and prediction.

[0027] Students' learning records include question sequences and interaction records. Each question is associated with one or more knowledge points, and there may be dependencies between knowledge points (for example, a knowledge point is a sub-knowledge point of another knowledge point).

[0028] In one embodiment, calculating the correlation between the sparse data space and different dense data spaces by weighted average difference of the topic embedding in step S2 and selecting the optimal dense data space for the sparse data space specifically includes: All the topic embeddings in the sparse data space or dense data space are subtracted pairwise to obtain difference embeddings, and the central characteristics of the sparse data space or dense data space are obtained by weighted average difference embedding. The correlation between the central characteristics of the sparse data space and the central characteristics of different dense data spaces is compared by cosine similarity, and the dense data space corresponding to the highest correlation is selected as the optimal dense data space of the sparse data space.

[0029] The central features are obtained through the above method. The reason for directly obtaining the central characteristics of the original data space is that when the difference is embedded in different data spaces, the central characteristics are more correlated and easier to migrate later.

[0030] The dense data space provides rich embedding information of questions and interactions. The knowledge state navigation module (KSN) learns these embeddings, identifies the differences in the sequence, and then extracts knowledge variables that can guide the learning process. The knowledge sequence analysis module (KSA) assists KSN by analyzing these knowledge variables to optimize and correct the knowledge tracking process. Through this collaboration, the system can more effectively understand and predict the changes in students' knowledge states.

[0031] Considering the characteristics of knowledge transfer, the present invention believes that the selection method of dense data space should meet two conditions: (1) a processed dense data space should be generated so that the correlation between the dense data spaces is greater than the correlation between the original feature spaces; (2) the correlation between the processed dense data space and the original feature space should be less than the correlation between the original feature spaces. Figure 3 As shown in (a) in the figure, after pre-training in different dense data spaces, we can embed any two topics ( and ) are added and subtracted to obtain the difference embedding space Common-diff and the aggregation embedding space Common-aggr. After weighted average embedding of the difference embedding space Common-diff and the aggregation embedding space Common-aggr, the cosine similarity between the three data spaces is calculated, as shown in Figure 3 As shown in (b) in . In some preferred embodiments, the difference embedding space Common-diff better satisfies the first condition. Figure 3 As shown in (c) in Figure 1, by comparing the similarity between the difference embedding space Common-diff and the aggregation embedding space Common-aggr, as well as the similarity between them and the corresponding original feature space, it can be found that the difference embedding space Common-diff better meets the second condition. Therefore, the difference embedding space can be used as the optimal dense data space.

[0032] In one embodiment, the sequence difference analysis of the topic embedding and the interaction embedding is performed to obtain the guiding knowledge variable, which specifically includes: In the optimal dense data space, the offset subtraction operation is performed on the question embedding sequence and the interaction embedding sequence to generate the question difference sequence and the interaction difference sequence; Input the question difference sequence into a first decoder and output question comprehension features; Input the interaction difference sequence into the second decoder and output the learning behavior feature; The question comprehension features are used as queries and keys, and the learning behavior features are used as values, which are input into the third decoder to generate guiding knowledge variables.

[0033] The knowledge state guides the process, which aims to process the difference information and provide the migration parameters for transfer learning. Figure 2As shown, the present invention performs a dislocation subtraction operation on the question embedding sequence and the interaction embedding sequence to obtain respective difference sequences. The knowledge tracking model of the present invention performs this operation on the question and interaction embedding respectively by subtracting the embedding at the previous moment from the embedding at the current moment, thereby obtaining the corresponding difference sequences. Logically and intuitively, it can be naturally understood that the information processed from the question embedding is information related to "understanding" the question, while the information processed from the interaction embedding is information related to "learning". In order to process these two types of information, the present invention adopts a first decoder and the second decoder To process these two types of information separately. The first decoder Output question understanding features As query and key, the second decoder outputs the learned behavior features As a value, input to the third decoder In order to obtain the guiding knowledge variable .

[0034] In dense data space, the parameters of the knowledge state guidance process are learnable, thereby obtaining information corresponding to the dense data space. In sparse data space, the parameters of the knowledge state guidance process are frozen and replaced in the sparse data space, thereby realizing knowledge transfer.

[0035] In one embodiment, the first decoder, the second decoder and the third decoder all adopt a monotonic multi-head attention mechanism.

[0036] In one embodiment, the method of modeling the student's knowledge state by using the questions and interaction records in the sparse data space to generate the student's knowledge state variable specifically includes: Modeling student interaction record sequences in sparse data space ; They represent the questions answered by the students at time t and the variables indicating whether the answers given by the students are correct. The interaction record sequence is input into the pre-trained knowledge tracking model to generate the student knowledge state variable, which represents the student's mastery of the relevant knowledge points.

[0037] In the sparse data space, the knowledge sequence analysis module (KSA) is used to model the students' interaction records, with the aim of extracting variables that can reflect the students' current knowledge level from the limited and incomplete learning data. KSA analyzes the interaction records between students and the learning system, identifies their learning behaviors and knowledge mastery status, and helps the education system to effectively evaluate and track students' learning progress in the case of incomplete data. This process enables the system to understand the students' knowledge status and provide personalized learning support.

[0038] In one embodiment, the classifier uses a multi-layer perceptron. The guiding knowledge variables provided by the knowledge state guidance process in the dense data space , and the student's knowledge state variables provided by the knowledge sequence analysis process in the sparse data space , add to get the fused variable , combined with the embedding of the next moment question , input into the multi-layer perceptron (MLP) for prediction, and output the probability of the student correctly answering the question at the next moment .

[0039] During the training process, the binary cross entropy loss function is used to optimize the knowledge tracking model to reduce the error between the predicted results and the actual labels. The entire process is completed alternately in dense data space and sparse data space, and finally realizes the migration of knowledge from dense data space to sparse data space, and significantly improves the knowledge tracking performance in sparse data space.

[0040] In some preferred embodiments, the present invention proposes a knowledge tracking method combined with knowledge migration, which migrates knowledge in dense data space to sparse data space by means of domain migration, thereby improving the performance of the knowledge tracking model in sparse data space. First, by calculating the cosine similarity in the dense data space, the most suitable dense data space is selected, and guidance information is extracted from it. Then, by freezing the parameters of the knowledge state guidance process and migrating it to the sparse data space, the model can be trained in a sparse data environment without manually adjusting the migration function. This method effectively solves the challenges of knowledge tracking in sparse data space, especially when the interaction records are short and the data is sparse, and provides an innovative solution to improve the prediction accuracy of the model in a sparse data environment.

[0041] The present invention demonstrates a complete technical process from data preprocessing, knowledge transfer to prediction through the above embodiments, which can be applied to intelligent learning evaluation and personalized recommendation in various educational scenarios.

[0042] In one embodiment, the present invention further discloses a migration knowledge tracking system in a sparse data scenario, comprising: Data space construction module: Encode the topics and interaction records in the sparse data set and multiple dense data sets into a sparse data space and multiple dense data spaces respectively. Both the sparse data space and the dense data space contain topic embedding and interaction embedding. Dense data space selection module: calculates the correlation between sparse data space and different dense data spaces through the weighted average difference of question embedding, and selects the optimal dense data space for the sparse data space; Knowledge tracking task module: The knowledge tracking task is performed in the dense data space through the knowledge state navigation module and the knowledge sequence analysis module; the knowledge state navigation module of the optimal dense data space is frozen, and the sequence difference analysis of the question embedding and the interaction embedding is performed to obtain the guiding knowledge variables; the knowledge sequence analysis module of the sparse data space is used to model the students' interaction records and generate the students' knowledge state variables; Transfer prediction module: After fusing the guiding knowledge variables and knowledge state variables, they are input into the classifier together with the embedding of the question at the next moment to predict the probability that the student will answer the question correctly at the next moment; Model training module: Cross-training is performed by freezing the parameters of the knowledge state navigation module in the optimal dense data space and migrating it to the sparse data space.

[0043] The system of the present invention corresponds to the method, and the specific embodiments applicable to the method are also applicable to the system. The specific embodiments of the method have been described in detail above and will not be repeated here.

[0044] In an exemplary embodiment, the present invention also provides a computer-readable storage medium including instructions, such as a memory including instructions, and the above instructions can be executed by a processor to perform the above method. The storage medium can be a computer-readable storage medium, for example, the computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0045] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention, and any reference numerals in the claims should not be regarded as limiting the claims involved.

[0046] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A migration knowledge tracking method in a sparse data scenario, characterized in that: The construction process of the adopted knowledge tracing model includes: Encode the topics and interaction records in the sparse data set and multiple dense data sets into a sparse data space and multiple dense data spaces respectively, and both the sparse data space and the dense data space contain topic embedding and interaction embedding; The correlation between the sparse data space and different dense data spaces is calculated through the weighted average difference of the question embedding, and the optimal dense data space is selected for the sparse data space; The knowledge tracking task is carried out in the optimal dense data space through the knowledge state navigation module and the knowledge sequence analysis module; the knowledge state navigation module of the optimal dense data space is frozen, and the sequence difference analysis of the question embedding and the interaction embedding is performed to obtain the guiding knowledge variables; the knowledge sequence analysis module of the sparse data space is used to model the students' interaction records and generate the students' knowledge state variables; After the guiding knowledge variable and the knowledge state variable are fused, they are input into the classifier together with the embedding of the question at the next moment to predict the probability that the student will answer the question correctly at the next moment. Cross-training is performed by freezing the parameters of the knowledge state navigation module in the optimal dense data space and migrating it to the sparse data space.

2. The migration knowledge tracking method in sparse data scenario according to claim 1 is characterized in that: The encoding of the topics and interaction records in the sparse data set and the plurality of dense data sets into a sparse data space and a plurality of dense data spaces, respectively, wherein both the sparse data space and the dense data space contain topic embedding and interaction embedding, specifically includes: Through the selected knowledge tracking model, the questions and interaction records in the sparse data set and dense data set are encoded to obtain sparse data space and dense data space; in the sparse data space and dense data space, each question and interaction record is represented as question embedding and interaction embedding respectively, which capture the characteristics of the question and the relevant information of the interaction with the students.

3. The migration knowledge tracking method in a sparse data scenario according to claim 1 is characterized in that: The weighted average difference of the embedded questions is used to calculate the correlation between the sparse data space and different dense data spaces, and the optimal dense data space is selected for the sparse data space, specifically including: All the topic embeddings in the sparse data space or dense data space are subtracted pairwise to obtain difference embeddings, and the central characteristics of the sparse data space or dense data space are obtained by weighted average difference embedding. The correlation between the central characteristics of the sparse data space and the central characteristics of different dense data spaces is compared by cosine similarity, and the dense data space corresponding to the highest correlation is selected as the optimal dense data space of the sparse data space.

4. The method for tracking migration knowledge in a sparse data scenario according to claim 1, characterized in that: The sequence difference analysis of the topic embedding and the interaction embedding is performed to obtain the guiding knowledge variables, which specifically include: In the optimal dense data space, the offset subtraction operation is performed on the question embedding sequence and the interaction embedding sequence to generate the question difference sequence and the interaction difference sequence; Input the question difference sequence into a first decoder and output question comprehension features; Input the interaction difference sequence into the second decoder and output the learning behavior feature; The question comprehension features are used as queries and keys, and the learning behavior features are used as values, which are input into the third decoder to generate guiding knowledge variables.

5. The method for tracking migration knowledge in a sparse data scenario according to claim 4, characterized in that: The first decoder, the second decoder and the third decoder all adopt a monotonic multi-head attention mechanism.

6. The method for tracking migration knowledge in a sparse data scenario according to claim 1, characterized in that: The knowledge sequence analysis module using the sparse data space models the student's interaction records and generates the student's knowledge state variables, specifically including: Modeling student interaction record sequences in sparse data space ; They represent the questions answered by the students at time t and the variables indicating whether the answers given by the students are correct. The interaction record sequence is input into the pre-trained knowledge tracking model to generate the student knowledge state variable.

7. The method for tracking migration knowledge in a sparse data scenario according to claim 1, characterized in that: The classifier adopts a multi-layer perceptron.

8. A migration knowledge tracking system in a sparse data scenario, characterized in that: include: Data space construction module: Encode the topics and interaction records in the sparse data set and multiple dense data sets into a sparse data space and multiple dense data spaces respectively. Both the sparse data space and the dense data space contain topic embedding and interaction embedding. Dense data space selection module: calculates the correlation between sparse data space and different dense data spaces through the weighted average difference of question embedding, and selects the optimal dense data space for the sparse data space; Knowledge tracking task module: The knowledge state navigation module and the knowledge sequence analysis module jointly perform knowledge tracking tasks in the optimal dense data space; freeze the knowledge state navigation module in the optimal dense data space, and perform sequence difference analysis on question embedding and interaction embedding to obtain guiding knowledge variables; The knowledge sequence analysis module of sparse data space is used to model the students' interaction records and generate students' knowledge state variables; Transfer prediction module: After fusing the guiding knowledge variables and knowledge state variables, they are input into the classifier together with the embedding of the question at the next moment to predict the probability that the student will answer the question correctly at the next moment; Model training module: Cross-training is performed by freezing the parameters of the knowledge state navigation module in the optimal dense data space and migrating it to the sparse data space.

9. The migration knowledge tracking system in a sparse data scenario according to claim 8, characterized in that: The dense data space selection module subtracts all the topic embeddings in the sparse data space or the dense data space from each other to obtain difference embeddings, obtains the central characteristics of the sparse data space or the dense data space through weighted average difference embeddings, uses cosine similarity to compare the correlation between the central characteristics of the sparse data space and the central characteristics of different dense data spaces, and selects the dense data space corresponding to the highest correlation as the optimal dense data space of the sparse data space.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images