Knowledge tracking method based on causal invariance decoupling sequence

Through causal attention mechanism and causal intervention operations, the reaction sequence is decoupled and causal associations are identified, and the problem of false associations in the existing technology is solved, and the accuracy and interpretability of knowledge tracking are improved.

CN118940796AActive Publication Date: 2024-11-12HUAZHONG NORMAL UNIV

Patent Information

Application Number
CN202410961044.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-12
Estimated Expiration
2044-07-17

AI Technical Summary

Technical Problem

Existing knowledge tracking methods ignore data selection bias, resulting in false association problems. The model lacks adaptability and accuracy when facing new models, and cannot accurately identify the answer subsequences in the reaction sequence that affect future performance.

Method used

Using a knowledge tracking method based on causal invariance decoupling sequences, decoupling the reaction sequences through causal attention mechanism and causal intervention operations, identify causal associations, remove redundant and irrelevant parts, and improve prediction accuracy. and explainability.

Benefits of technology

Accurate identification of causal associations in learner response sequences is achieved, the accuracy and interpretability of knowledge tracking predictions are improved, and the adaptability and stability of the model are enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940796A_ABST
    Figure CN118940796A_ABST
Patent Text Reader

Abstract

The invention provides a knowledge tracking method based on a causal invariance decoupling sequence for solving the problem of false association caused by neglecting data selection bias in an existing knowledge tracking method, and the method comprises the steps: decoupling a reaction sequence into a cause subsequence and a trivial subsequence through a designed causal attention mechanism and causal intervention operation; therefore, the answer subsequences in causal association with the answer expressions are found, redundant and irrelevant answers are removed, the accuracy of the knowledge tracking prediction task is improved, and the interpretability of the knowledge tracking method is improved through the causal analysis technology. According to the method, the simplification of redundant data in the sequence is realized, the accuracy of the prediction task is improved, and the interpretability of the knowledge tracking method is greatly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the fields of educational data mining and knowledge tracking, and in particular to a method for dynamically tracking the evolution of a learner's knowledge state. Technical Background

[0002] Knowledge Tracing (KT) is a key method to achieve personalized learning. It predicts learners' future performance by analyzing their question-answering behavior, and dynamically tracks the evolution of knowledge status throughout the learning process. It provides support for learning resource recommendations and adaptive learning, and promotes the development of personalized education.

[0003] Deep Knowledge Tracing (DKT), proposed in 2015, is a groundbreaking work in the field of knowledge tracing. It pioneered the use of recurrent neural networks to encode reaction sequences to obtain knowledge states. Following DKT, a series of classic works have emerged, such as dynamic key-value memory networks and attention knowledge tracing. The former uses memory networks to store and dynamically update knowledge states, while the latter uses attention mechanisms to retrieve knowledge states from reaction sequences. In recent years, with the further development of deep learning, more excellent knowledge tracing methods have emerged. On the one hand, researchers continue to improve the feature extraction capabilities of neural networks and incorporate more advanced deep learning technologies, such as heterogeneous graph neural networks, contrastive learning, and multi-task learning. On the other hand, researchers continue to deepen their understanding of the evolution of knowledge states and incorporate more educational principles, such as modeling the laws of learning and forgetting, the influence of sequence time characteristics, the relationship between question difficulty and answering performance, and the relationship between response answering speed and knowledge states.

[0004] However, while current knowledge tracking methods have made significant progress in understanding and assessing students' knowledge mastery, they often ignore the impact of data selection bias. Specifically, the model may over-rely on the statistical properties of test questions, such as co-occurrence frequency, rather than the true impact on the response results, mistakenly believing that certain test questions are very important, thereby leading to selection bias. This reliance on statistical properties essentially represents a shortcut learning strategy, which can improve the prediction performance on test questions with similar patterns, but limits the adaptability and accuracy of the model when facing new patterns, leads to spurious correlations, and inhibits the exploration of causal relationships between response sequences and future performance.

[0005] Therefore, we use deep learning technology to model the learners' response data when answering test questions, explore a knowledge tracking method based on causal analysis technology, accurately identify the responses in the response sequence that really affect future performance, and distinguish them from irrelevant parts, thereby revealing the causal relationship in the response sequence and improving the accuracy and interpretability of the model, which has important research significance and application value. Summary of the invention

[0006] The purpose of the present invention is to propose a knowledge tracking method based on causal invariance decoupling sequence to address the problem of false correlation caused by existing knowledge tracking methods ignoring data selection bias. Through the causal attention mechanism and causal intervention operation designed by us, the response sequence is decoupled into causal sub-sequences and trivial sub-sequences, so as to find the answer sub-sequences that are causally related to the answer performance, remove redundant and irrelevant answers, improve the accuracy of knowledge tracking prediction tasks, and improve the interpretability of the knowledge tracking method through causal analysis technology.

[0007] To achieve the purpose of the invention, the present invention adopts the following technical solution: a knowledge tracking method based on causal invariance decoupling sequence, comprising:

[0008] Firstly, a knowledge tracking model based on causal invariance decoupled sequence is constructed and trained. The knowledge tracking model includes a reaction sequence encoder, a causal identifier based on attention mechanism, a sequence causal intervener and a future performance predictor. The total loss of the model is obtained by multi-task learning method, and finally the trained model is used to realize knowledge tracking.

[0009] The response sequence encoder is used to obtain the response sequence and the test question sequence, and the long short-term memory network is used as an encoder to encode the response sequence and the test question sequence to obtain the representation of the response sequence and the test question sequence;

[0010] Among them, the learner at time t and a question q t The interaction of the two forms a reaction unit x t =(q t ,c t ,r t ), each unit contains question q t 、The knowledge concepts involved t and the reaction result r t , the reaction sequence is composed of multiple reaction units combined in the order of answering;

[0011] The causal identifier based on the attention mechanism is used to construct a question-concept graph using the response sequence, and encode the question-concept graph through a graph convolutional network to obtain node representations of questions and concepts. The causal attention mechanism is used to identify the relationship between future questions and response sequences from the perspective of time and knowledge association, obtain the time association weight and knowledge association weight, and finally calculate the causal score and trivial score based on the time association weight and knowledge association weight;

[0012] The sequence causal interventioner is used to decompose the original complete response sequence into two parts: a causal subsequence and a trivial subsequence by using the causal score and the trivial score. The intervention sequence is generated by performing a causal intervention operation on the trivial subsequence and combining it with the causal subsequence. Finally, the encoder and pooling operations are used to obtain the intervention sequence representation and the trivial subsequence representation.

[0013] The future performance predictor is used to set the same future performance prediction task for different intervention sequence representations to discover the invariant and stable causal relationships therein. For trivial subsequence representations, a problem difficulty prediction task is introduced to eliminate prediction shortcuts, and the weights of the two tasks are balanced and optimized through multi-task learning to achieve the prediction of learners' future performance and the difficulty level of future problems.

[0014] Furthermore, the process of encoding the reaction sequence is as follows:

[0015] Feature embedding: Use random embedding technology to map the elements in the reaction unit into a high-dimensional vector space to obtain an embedded representation of the reaction sequence.

[0016] X T =[x1,x2,…,x t ,…,x T ],x t =q t +c t +r t

[0017] in represents the embedded representation of the reaction sequence, T represents the total length of the reaction sequence, and the embedded representations of the question, knowledge concept and reaction are denoted as q t ,c t , where d is the embedding dimension, represents the reaction unit representation at time t;

[0018] Response sequence encoding: Use the long short-term memory network as an encoder to model the knowledge acquisition and forgetting process in students’ learning, receive the embedded representation of students’ historical response sequences and encode them to obtain the potential knowledge state representation;

[0019]

[0020] in represents the potential knowledge state of the reaction unit at time t, and Represents the cell state, which transmits information between different moments; represents the output gate, represents the forget gate, represents the input gate, represents the candidate cell state, and thus the overall characterization of the reaction sequence is as follows:

[0021] H T =LSTM X (X T )=[h1,h2,…,h t ,…,h T ]

[0022] in Each row in represents the potential knowledge state of the corresponding reaction unit at each moment. The LSTM network for reaction sequence encoding is formally defined as LSTM X .

[0023] Furthermore, the process of encoding the test question sequence in step (1) is as follows:

[0024] The embedding representation of the test sequence is as follows:

[0025] O T+1 =Q T+1 +C T+1 =[(q1+c1),(q2+c2),…,(q t +c t ),…,(q T+1 +c T+1 )]

[0026] in represents the test question sequence embedding representation, and Represent the problem and concept sequence respectively;

[0027] The embedded representation of the input test sequence is then encoded using the LSTM network as follows:

[0028] E T+1 =LSTM E (O T+1 )=[e1,e2,…,e t+1 ,…,e T+1 ]

[0029] in Each row in represents the representation of the test question to be answered. Corresponding to the question at time t+1, the LSTM used to encode the question sequence is denoted as LSTM E , whose structure is similar to LSTM X The same, but the parameters are not shared.

[0030] Furthermore, the problem concept graph constructed based on the reaction sequence in step (2) is represented as Node Set It consists of the questions and knowledge concepts involved in the reaction sequence. The same questions or concepts at different times are regarded as different nodes. First, the graph node representation is initialized. Specifically, it is defined first:

[0031]

[0032] The edge set ε is bidirectional, including sequence relationship edges and question-concept edges. The sequence relationship edges are added between the nodes of adjacent reactions, and the question-concept edges are added between the corresponding nodes through the correspondence between the questions and the knowledge concepts.

[0033] For graph network updates, GCN is used to aggregate v n The neighbor nodes of the next layer are obtained, and the update formula is as follows:

[0034]

[0035] in represents the updated node representation of the nth node in the l+1th layer, in is the adjacency matrix A n The nth row in represents the node v n Relationships with other nodes; represents the embedding matrix of all nodes in layer l, represents the weight matrix of the lth layer, Dropout represents regularization, and ReLu is the activation function; after encoding the question-concept graph, the output representation of each node is expressed as Where L is the number of stacked GCN layers.

[0036] Furthermore, the specific implementation method of obtaining the time association weight and the knowledge association weight in step (2) is as follows:

[0037] First, from the perspective of time association, the test question at time t+1 is represented as e t+1 Set as the query vector, set the reaction unit representation containing time information as the key vector, and use scaled dot product attention. The weight between the unit at t+1 and η is calculated as follows:

[0038]

[0039] in, represents the representation of the future test question at time t+1, represents the reaction unit representation at time η, θ t+1,η Represents the time-related weight; weight matrix and Linear transformations used in attention mechanisms, is the scaling factor;

[0040] Secondly, from the perspective of knowledge association, we use the node representation after graph encoding to calculate the association between reaction units and represent the problem at time t+1 and conceptual representation Set it as the query vector, set the representation of other units in the sequence as the key vector, and the weight calculation is as follows:

[0041]

[0042] where ω t+1,η represents the knowledge association weight between the test questions at time t+1 in the future and other test questions at time η, ⊙ represents the Hadamard product, and multiplication is performed on the opposite elements; is the problem and concept representation after graph encoding, the weight matrix Used for linear transformations.

[0043] Furthermore, the specific implementation method of obtaining the causal score and the trivial score in step (2) is as follows;

[0044] The causal score is calculated as follows:

[0045]

[0046] in represents the causal score between the test question at the future time t+1 and the response unit at the time η, and They represent the time association weight and knowledge association weight used for causal score calculation respectively; is a masking matrix used to implement timing masking, with the upper triangular elements set to -∞ and the rest set to 0. t+1,μ is the masking value corresponding to t+1 and μ;

[0047] The trivial score is calculated as follows:

[0048]

[0049] in represents a trivial fraction, denote the time association weight and knowledge association weight used for trivial score calculation, respectively, and σ(·) denotes the sigmoid activation function;

[0050] Furthermore, in order to completely decompose the reaction sequence into causal and trivial subsequences, the scores of continuous values ​​are converted to discrete values. This is achieved using the gumbel-softmax function as shown below:

[0051]

[0052] The gumbel-softmax function converts the continuous input vector into a discrete output vector while allowing the gradient to back propagate. In {0,1}, therefore

[0053] Finally, the discretized attention weights are again time-masked to ensure that no information leakage will occur in the future; specifically, for the test questions at the future time t+1, if η ≥ t+1, then Set to 0, so the attention score of the test at time t+1 in the entire sequence is as follows:

[0054]

[0055] in Represent the final causal score and trivial score of the entire sequence respectively; in addition, the sum of the causal score and trivial score matrices formed by all future test questions is a unit lower triangular matrix, in which all elements on the diagonal and below the diagonal are 1, and the other elements are 0, expressed as in

[0056] Furthermore, the causal score and trivial score are multiplied with the entire reaction sequence to extract the corresponding causal subsequence and trivial subsequence, as shown below:

[0057]

[0058] in Denote the causal subsequence and trivial subsequence corresponding to time t+1, respectively. Each row in the matrix represents the embedding of a reaction unit. The non-causal or non-trivial part is filled with a zero vector through the Hadamard product. Thus, the reaction sequence is decoupled into two parts, and there is no overlap between the two subsequences, that is,

[0059] Will After intervention t+1 The conditional probability is expressed as follows:

[0060]

[0061] in Represents Zt+1 The set of all possible values, do(·) represents the Do operator;

[0062] This is approximated by considering a set of operations applied to trivial subsequences, resulting in a fixed number of intervening sequences, thus transforming it into the following approximation:

[0063]

[0064] where c t+1 Represents a trivial subsequence After the operation, the A collection,consists of causal subsequences with different contexts;

[0065] Next, for the trivial subsequence Perform specific intervention operations to obtain the intervention sequence set c t+1 as follows:

[0066]

[0067] in represents an intervention sequence, oper(·) represents three different intervention operations, including removal (rem), replacement (repl), and inversion (inve);

[0068] (3-3) Intervention sequence representation

[0069] Finally, each intervention sequence is encoded to obtain the corresponding representation. The LSTM encoder is applied to each intervention sequence to capture the sequence dependency and encode the sequence into a fixed-length representation. The overall representation of the sequence is then obtained through an average pooling operation. The formula is as follows:

[0070]

[0071] in Represents intervention sequence The average pooling representation of t+1 represents a set of representations; in particular, It can also be seen as a causal subsequence Representation; LSTM X represents the response sequence encoder;

[0072] The representation of a trivial subsequence is calculated as follows:

[0073]

[0074] in represents a trivial subsequence representation, and rem(·) represents a removal operation.

[0075] Furthermore, the future performance prediction task specifically includes:

[0076] Based on the causal subsequence, trivial subsequence and future test representation, the conditional probability of future performance is defined as follows:

[0077]

[0078] This shows that future performance t+1 is affected by the causal subsequence and the representation of future questions, while the trivial subsequence has no direct causal effect on prediction; therefore, different intervention sequences can be used for the same future performance prediction task, and their prediction results are roughly similar;

[0079]

[0080] in represents the predicted probability of the intervention sequence, and cat[·] represents the concatenation along the feature dimension. These are trainable parameters that are not shared across operations.

[0081] Finally, the binary cross entropy loss function is used to measure the error between the predicted probability and the actual outcome. The loss for each intervention sequence is defined as follows:

[0082]

[0083] where r t+1 ∈{0,1} represents the actual response result at time t+1. For different intervention sequences, the loss is calculated separately. Enables the model to extract stable and invariant causal relationships from different sequences.

[0084] Furthermore, the difficulty prediction task specifically includes:

[0085] Discretize continuous values ​​into difficulty levels and use question difficulty as the prediction target for trivial subsequences; specifically, first use all answer records to calculate the average correct rate of questions, then sort the correct rates and divide them into nd difficulty levels, each level contains the same number of questions, calculated as follows:

[0086]

[0087] where d t+1 represents the difficulty level of the question to be answered at time t+1, f is the mapping function from accuracy to difficulty level, δ(·) is the indicator function, which returns 1 if the condition is true, otherwise it returns 0; q ij and r ij Represent a question and answer in all records respectively; then, use trivial subsequence representation To predict the difficulty level of future questions, as follows:

[0088]

[0089] in represents the predicted probability of the difficulty of the question, and is a trainable parameter, cat[·] represents the cascade operation along the feature dimension; finally, binary cross entropy is used to measure the error between the predicted probability and the actual difficulty level of the trivial subsequence:

[0090]

[0091] in is the loss value of the problem difficulty prediction task, I(d t+1 ) indicates the difficulty d t+1 Whether it belongs to the corresponding difficulty category.

[0092] Furthermore, the losses of the future performance prediction task and the problem difficulty prediction task are combined to calculate the overall loss, which is defined as follows:

[0093]

[0094] where λ1, λ2, λ3, and λ4 are weighting coefficients used to balance the contribution of each loss term, correspond to the losses of deletion, substitution and reversal operations respectively, and is the predicted probability of the problem difficulty.

[0095] Compared with the prior art, the present invention has the following obvious outstanding substantive features and remarkable technological progress:

[0096] (1) The present invention proposes a novel causal analysis method suitable for reaction sequences in knowledge tracking. The method distinguishes causal subsequences from trivial subsequences through an identifier, generates intervention sequences based on causal invariance through an intervener, and sets multiple prediction targets in the predictor to discover stable and invariant causal relationships.

[0097] (2) The present invention proposes a causal attention mechanism that estimates the causal and trivial scores between future questions and the representation of each reaction unit by combining time and knowledge associations. The score calculation of this attention mechanism is more accurate and reasonable, and there is no overlap between the causal and trivial scores obtained.

[0098] (3) The present invention proposes three intervention operations suitable for reaction sequences. These intervention operations are guided by causal invariance, conform to the characteristics of reaction sequences and our common sense, and can be efficiently implemented through algorithms.

[0099] (4) The present invention proposes a knowledge tracking method based on causal invariance decoupled sequence, which is carried out from the perspective of causal analysis. By decoupling the sequence into causal sub-sequences and trivial sub-sequences, it realizes the simplification of redundant data in the sequence, improves the accuracy of the prediction task, and greatly improves the interpretability of the knowledge tracking method. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 This is a framework diagram of the knowledge tracking method based on causal invariance decoupling sequence.

[0101] Figure 2 This is the structural causal model diagram of CIKT. DETAILED DESCRIPTION

[0102] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0103] The theoretical basis for the implementation of the present invention is first described below:

[0104] First, let’s talk about the problem definition: In knowledge tracing, the learner is at time t and has a question q t The interaction of the two forms a reaction unit x t =(q t ,c t ,r t ). Each unit contains questions q t 、The knowledge concepts involved t and the reaction result r t Questions and knowledge concepts are usually represented by indexes (i.e., digital numbers), while results are represented by 0 and 1, representing incorrect and correct responses, respectively. A response sequence is composed of multiple such response units in the order of answering. As of time t, a series of response units answered by the learner form a response sequence X t ={(q1,c1,r1),(q2,c2,r2),...,(q t ,c t ,r t )}, accordingly, the question sequence and concept sequence can be expressed as Q t ={q1,q2,...,q t} and C t ={c1,c2,...,c t The goal of knowledge tracking is to infer the latent knowledge state based on the learner’s performance and to track the dynamic evolution of the knowledge state through the response sequence. The total number of response units for each student is T+1, and the KT prediction task is defined as: for each t∈[1,T], based on the corresponding response sequence X t={(q1,c1,r1),(q2,c2,r2),...,(q t ,c t ,r t )}, predict the learner in q t+1 and c t+1 The reaction on t+1 , denoted as P(r t+1 =1|X t ,q t+1 ,c t+1 ).

[0105] Structural causal model analysis:

[0106] like Figure 2 As shown in Figure 1, a structural causal model for knowledge tracing is constructed. By analyzing the key elements in the KT task, it can be found that the complete reaction sequence S includes a causal subsequence C and a trivial subsequence T. The causal subsequence C is retrieved from S by using the future question F. Therefore, the causal subsequence C is oriented to a specific future question F, and corresponds to it one by one, and has a causal relationship with the future performance Y. However, the trivial subsequence T is affected by the data selection bias and has a false correlation with Y. In other words, given F and C, Y and T are independent, that is, Y⊥T∣C,F. Based on causal invariance, as long as the causal subsequence C remains unchanged, the causal relationship between the sequence as a whole and the future performance Y should remain stable regardless of how the trivial subsequence T changes. In addition, it can be found that the trivial subsequence T opens a backdoor path through the causal chain C←S→T→H→Y. Therefore, the most direct and effective method is to change the prediction target of T and link it with the problem difficulty D, thereby changing the impact of this path. Through the structural causal model, we analyzed the relationship between elements in knowledge tracking from a causal perspective, providing guidance for subsequent model design and step implementation.

[0107] This paper proposes a novel knowledge tracing method based on disentangling response sequences with causal invariance (CIKT). Figure 1As shown in the figure, the CIKT model consists of four modules: a reaction sequence encoder, a causal identifier based on an attention mechanism, a sequence causal intervener, and a future performance predictor. Specifically, when predicting the learner's future performance, the encoder first uses LSTM to obtain the representation of the historical reaction sequence and the future test sequence; then, the causal identifier enhances the attention score calculation between the reaction sequence and the future question by combining time and knowledge association, thereby facilitating the decomposition of the sequence into causal and trivial subsequences; subsequently, the intervener uses intervention strategies including removal, replacement, and reversal based on the principle of causal invariance to generate intervention sequences, and uses the reaction sequence encoder to obtain its corresponding representation; finally, the predictor sets two prediction targets for the intervention sequence, including future performance prediction and problem difficulty prediction, driving CIKT to explore the causal relationship between the reaction sequence and future performance. Specifically include:

[0108] Firstly, a knowledge tracking model based on causal invariance decoupled sequence is constructed and trained. The knowledge tracking model includes a reaction sequence encoder, a causal identifier based on attention mechanism, a sequence causal intervener and a future performance predictor. The total loss of the model is obtained by multi-task learning method, and finally the trained model is used to realize knowledge tracking.

[0109] The response sequence encoder is used to obtain the response sequence and the test question sequence, and the long short-term memory network is used as an encoder to encode the response sequence and the test question sequence to obtain the representation of the response sequence and the test question sequence;

[0110] The causal identifier based on the attention mechanism is used to construct a question-concept graph using the response sequence, and encode the question-concept graph through a graph convolutional network to obtain node representations of questions and concepts. The causal attention mechanism is used to identify the relationship between future questions and response sequences from the perspective of time and knowledge association, obtain the time association weight and knowledge association weight, and finally calculate the causal score and trivial score based on the time association weight and knowledge association weight;

[0111] The sequence causal interventioner is used to decompose the original complete response sequence into two parts: a causal subsequence and a trivial subsequence by using the causal score and the trivial score. The intervention sequence is generated by performing a causal intervention operation on the trivial subsequence and combining it with the causal subsequence. Finally, the encoder and pooling operations are used to obtain the intervention sequence representation and the trivial subsequence representation.

[0112] The future performance predictor is used to set the same future performance prediction task for different intervening sequence representations to discover the invariant and stable causal relationships therein. For trivial subsequence representations, a difficulty prediction task is introduced to eliminate prediction shortcuts, and the weights of the two tasks are balanced and optimized through multi-task learning to achieve the prediction of learners' future performance and the difficulty level of future problems.

[0113] (1) Reaction sequence coding

[0114] In order to decouple the reaction sequence, we first need to obtain the representation of the reaction sequence and the test sequence. For the encoder network structure, we use the Long Short-Term Memory (LSTM) network for sequence encoding, which is widely used in sequence tasks and can effectively capture the temporal information of the reaction unit. The steps mainly include feature embedding, reaction sequence encoding, and test sequence encoding.

[0115] (1-1) Feature Embedding

[0116] In the previous problem definition, the learner's response unit at time t is denoted by x t =(q t ,c t ,r t ). We use random embedding techniques to map the elements in the response unit into a high-dimensional vector space. The embedded representations of questions, knowledge concepts, and responses are denoted as q t ,c t , Where d is the embedding dimension. We get the embedding representation of each reaction unit by adding the corresponding reaction elements to form a reaction sequence as follows:

[0117] X T =[x1,x2,…,x t ,…,x T ],x t =q t +c t +r t

[0118] in represents the embedded representation of the reaction sequence, T represents the total length of the reaction sequence, represents the reaction unit representation at time t. The question sequence involves questions to be answered in the future, and its embedding representation is as follows:

[0119] O T+1 =Q T+1 +C T+1 =[(q1+c1),(q2+c2),…,(q t +c t ),…,(q T+1 +c T+1 )]

[0120] in represents the test question sequence embedding representation, Represent questions and concept sequences respectively.

[0121] (1-2) Reaction sequence coding

[0122] In order to predict future performance, we need to understand the student's past performance on the problem. Therefore, we set up a response sequence encoder to receive the student's historical response sequence and encode the potential knowledge state. The knowledge state at each moment is affected by the previous state, so we use LSTM to model the knowledge acquisition and forgetting process in student learning. The knowledge state at each moment is calculated as follows:

[0123]

[0124] in represents the potential knowledge state of the reaction unit at time t, and Represents the cell state, which conveys information between different moments. represents the output gate, represents the forget gate, represents the input gate, represents the candidate cell state. Thus, the overall characterization of the reaction sequence is as follows:

[0125] H T =LSTM X (X T )=[h1,h2,…,h t ,…,h T ]

[0126] in Each row in represents the potential knowledge state of the corresponding reaction unit at each moment. We formally define the LSTM network for reaction sequence encoding as LSTM X .

[0127] (1-3) Question sequence coding

[0128] Future performance depends not only on past responses but also on future questions to be answered. Therefore, we use a second encoder to obtain representations of future questions. Similar to the response sequence encoder, we input the embedding representation of the question sequence and then encode it using an LSTM network as follows:

[0129] E T+1 =LSTM E (O T+1 )=[e1,e2,…,e t+1 ,…,e T+1 ]

[0130] in Each row in represents the representation of the test question to be answered. The LSTM for question sequence encoding is denoted as LSTME , although its structure is similar to LSTM X Similar, but the parameters are not shared.

[0131] (2) Score calculation based on causal attention

[0132] Existing KT methods usually use the entire reaction sequence to predict future performance. Although this provides as much information as possible for prediction, the irrelevant information contained in it actually interferes with the prediction to a certain extent, such as spurious relationships caused by data bias. The model may mistakenly attribute a significant impact on answer performance to a specific question based only on superficial statistical properties, thus failing to deeply explore causal relationships. Therefore, we intend to actively use the association of questions in terms of knowledge concepts to help decompose the reaction sequence into causal and trivial subsequences. Specifically, we first use a graph convolutional network (GCN) to mine the association between questions through a question-concept graph, and then incorporate knowledge associations into causal attention calculations to obtain the causal score and trivial score of the reaction sequence. The specific steps include: question-concept graph encoding, causal attention calculation, and causal score calculation.

[0133] (2-1) Question-concept map coding

[0134] In order to exploit the graph network structure to mine associations, we constructed a question concept graph G based on the reaction sequence. QC =(v,ε). The node set v consists of the questions and knowledge concepts involved in the sequence, and the same questions or concepts at different times are regarded as different nodes. We use the feature embedding module in (1-1) to initialize the graph node representation. Specifically, we define:

[0135]

[0136] The edge set ε is bidirectional, including sequence relationship edges and question-concept edges. Sequence relationship edges are added between nodes of adjacent reactions, such as {q1,q2}, {c1,c2}, {q1,c2}, {c1,q2}. Question-concept edges are added between corresponding nodes through the correspondence between questions and knowledge concepts. For example, if q1 and q5 examine the same knowledge concept, there are edge relations {q1,c1}, {q5,c5}, {q1,c5} and {c1,q5}. For graph network updates, we use GCN aggregation v n The neighbor nodes of the next layer are obtained, and the update formula is as follows:

[0137]

[0138] in represents the updated node representation of the nth node in the l+1th layer, in is the adjacency matrix A n The nth row in represents the node v n Relationships with other nodes. represents the embedding matrix of all nodes in layer l, represents the weight matrix of the lth layer. Dropout is used for regularization to improve the generalization ability of the model. ReLu is a common activation function used for nonlinear transformation in neural networks. After encoding the question-concept graph, the output representation of each node is expressed as Where L is the number of stacked GCN layers.

[0139] (2-2) Causal Attention Calculation

[0140] When predicting future performance, we use future test questions to retrieve in the response sequence. To this end, we design a causal attention mechanism to identify the relationship between future questions and the response sequence from the perspective of time and knowledge association, and obtain causal and trivial scores for subsequent decoupling.

[0141] From the perspective of temporal association, learners’ performance on future questions depends on their performance on past response units. Therefore, we represent the question at time t+1 as e t+1 Set as the query vector, and set the reaction unit representation containing time information as the key vector. We use scaled dot product attention, and the weight between the unit at t+1 and η is calculated as follows:

[0142]

[0143] in, represents the representation of the future test question at time t+1, represents the reaction unit representation at time η, θ t+1,η Represents the temporal association weight between them. The weight matrix Linear transformations used in attention mechanisms, is the scaling factor.

[0144] From the perspective of knowledge association, response units that are closely connected in knowledge are more likely to show causal relationships. Therefore, we use the node representation after graph encoding to calculate the association between response units. We represent the problem at time t+1 as and conceptual representation Set it as the query vector, and set the representation of other units in the sequence as the key vector. The weight calculation is as follows:

[0145]

[0146] where ω t+1,η Represents the knowledge association weight between the test questions at time t+1 in the future and other test questions at time η. ⊙ represents the Hadamard product, which performs multiplication on the corresponding elements. It is the problem and concept representation after the graph encoding in step (2-1). The weight matrix The same is true for linear transformations.

[0147] (2-3) Causal score calculation

[0148] We combine the weights obtained from the two perspectives to obtain the final causal score and trivial score. For the causal score, we expect that the more relevant the future test questions are to the reaction unit in the sequence, the higher their score will be. The causal score is calculated as follows:

[0149]

[0150] in Represents the causal score between the test question at the future time t+1 and the response unit at time η. and Calculated by the two formulas in step (2-2), they respectively represent the time association weight and knowledge association weight used for causal score calculation. is a masking matrix used to implement timing masking, with its upper triangular elements set to -∞ and the rest set to 0. t+1,η is the masked value corresponding to t+1 and η. In contrast, for the triviality score, we want the response units with lower correlation to get higher triviality scores. Therefore, we use the negative sign and sigmoid function when calculating the triviality score, as shown below:

[0151]

[0152] in Represents a trivial fraction. The two formulas in step (2-2) are also used to calculate the time association weight and knowledge association weight for trivial score calculation, but different parameters W are used. Q and W K , these parameters are not shared. σ(·) represents the sigmoid activation function.

[0153] Furthermore, in order to fully decompose the reaction sequence into causal and trivial subsequences, we need to convert the scores of continuous values ​​into discrete values. We use the gumbel-softmax function to achieve this as follows:

[0154]

[0155] The gumbel-softmax function can convert continuous input vectors into discrete output vectors while allowing gradient backpropagation. and In {0,1}, therefore

[0156] Finally, we need to perform temporal masking on the discretized attention weights again to ensure that no information leakage will occur in the future. Specifically, for the test questions at the future time t+1, if η ≥ t+1, then Set to 0. Therefore, the attention score of the test question at time t+1 in the entire sequence is as follows:

[0157]

[0158] in Represent the final causal score and trivial score of the entire sequence respectively. In addition, the sum of the causal score and trivial score matrices formed by all future test questions is a unit lower triangular matrix, in which all elements on the diagonal and below the diagonal are 1, and the other elements are 0, which is expressed as in In summary, through this step, we obtain the causal score and the trivial score, which will be used for subsequent sequence decoupling.

[0159] (3) Sequential Causal Intervention

[0160] In the KT task, we aim to predict students' future performance based on their past responses. However, not every past response has a causal relationship with future performance. Due to data selection bias, some units may show spurious correlations. Based on the principle of causal invariance, if the response unit has a stable causal relationship with future performance, this relationship should remain unchanged in different environments. Therefore, for each future test question, CIKT first decomposes the response sequence into causal subsequences and trivial subsequences. Subsequently, CIKT performs intervention operations to delete, replace, and reverse elements in trivial subsequences to change the context of the causal subsequence. A variety of intervention sequences are generated through these interventions, which are used to discover causal relationships that remain unchanged in different contexts. The specific steps include: sequence decoupling, causal intervention, and intervention sequence representation.

[0161] (3-1) Sequence Decoupling

[0162] At different moments, the questions to be answered and the corresponding reaction sequences are different. Therefore, for each future question, there is a corresponding causal subsequence and a trivial subsequence. Each causal score and trivial score is either 0 or 1, which actually acts as a masking function. We use these scores to determine which subsequence each reaction unit belongs to. Specifically, we multiply these scores with the entire reaction sequence to extract the corresponding subsequence, as follows:

[0163]

[0164]

[0165] in Denote the causal subsequence and trivial subsequence corresponding to time t+1, respectively. Each row in the matrix represents the embedding of a reaction unit, and the non-causal or non-trivial part is filled with zero vectors through the Hadamard product. Thus, the reaction sequence is decoupled into two parts, and there is no overlap between the two subsequences, that is,

[0166] (3-2) Causal Intervention

[0167] Our goal is to estimate the causal subsequence Future performance t+1 To avoid the confounding effects of responses that are unrelated to future performance, we adjust the trivial subsequence to change the environment of the causal subsequence. According to the backdoor adjustment theory, we will After intervention t+1 The conditional probability is expressed as follows:

[0168]

[0169] in Represents Z t+1 The set of all possible values ​​that Z can take (i.e., irrelevant responses that cause confusion). do(·) represents the Do operator, which intervenes on the variable. This equation is derived from the backdoor adjustment formula, but in practice, due to computational limitations, it is usually not possible to consider Z in detail. t+1 all possible values ​​of . We therefore approximate this by considering a set of operations applied to trivial subsequences, generating a fixed number of intervening sequences. We thus transform this into the following approximation:

[0170]

[0171] where c t+1 Represents a trivial subsequence After the operation, the A collection of causal subsequences with different contexts.

[0172] Next, we have the trivial subsequence Perform specific intervention operations to generate multiple sequences. Specifically, considering the characteristics of the reaction sequence, we designed three types of intervention operations: removal (rem), replacement (repl), and inversion (inve). These operations are consistent with common sense cognition of reaction patterns and can be efficiently implemented through simple algorithms. The set of intervention sequences c t+1 as follows:

[0173]

[0174] in represents the intervention sequence, and oper(·) represents three different intervention operations. These three operations are introduced as follows:

[0175] Remove (rem). Since trivial subsequences have no causal relationship with future performance, discarding them will not change the causal relationship. Therefore, we directly remove trivial subsequences, making Only the cause and effect part is included

[0176] Replace (repl). Students tend to perform similarly on similar problems. For each problem q in the trivial subsequence t , we find problems with the same knowledge concept, randomly select one with replacement, and then combine it with the causal subsequence.

[0177] If the reaction unit has no causal relationship with future performance, then it does not matter whether the result is correct. Therefore, we convert each reaction r in the trivial subsequence t Reversal, reversing correct and incorrect responses.

[0178] It is worth noting that since trivial subsequences have no causal relationship with future performance, although these operations will change the context around the causal subsequence, they will not affect the inherent causal relationship between the sequence as a whole and the answer performance.

[0179] (3-3) Intervention sequence representation

[0180] Finally, we encode each intervention sequence to obtain the corresponding representation. By applying the LSTM encoder to each intervention sequence, we can capture the sequence dependencies and encode the sequence into a fixed-length representation, and then obtain the overall representation of the sequence through an average pooling operation. The formula is as follows:

[0181]

[0182] in Represents intervention sequence The average pooling representation oft+1 represents a set of representations. In particular, It can also be seen as a causal subsequence Representation of LSTM X Shares parameters with the reaction sequence encoder described in steps (1-2).

[0183] In addition, to further facilitate the discovery of causal relationships and avoid prediction shortcuts, we also consider the case where the causal subsequence is deleted, leaving only the trivial subsequence. The representation of the trivial subsequence is calculated as follows:

[0184]

[0185] in represents a trivial subsequence representation, and rem(·) represents a removal operation.

[0186] In summary, in the Sequential Causal Intervener, we generate a set of intervention sequences and obtain their corresponding representations through different intervention operations. These sequences are used in the next step to discover the invariant causal relationship between the sequence and the future representation.

[0187] (4) Future performance forecast

[0188] In the Predictor module, our goal is to predict future performance based on the intervention sequence generated in the Causal Intervenor module. Specifically, we Set up the same future performance prediction task to discover the invariant causal relationships and Set up the problem difficulty prediction task to eliminate prediction shortcuts. And balance and optimize the weights of these tasks through multi-task learning. The specific steps include: future performance prediction, problem difficulty prediction, and total loss calculation.

[0189] (4-1) Future performance forecast

[0190] For c t+1 In the intervention sequence, although the trivial subsequence is removed, replaced or reversed, the core causal subsequence remains unchanged, thus preserving its causal relationship. Based on the causal subsequence, trivial subsequence and future test representation, the conditional probability of future performance is defined as follows:

[0191]

[0192] Future performance can be discovered t+1 is mainly affected by the causal subsequence and the representation of future questions, while the trivial subsequence has no direct causal effect on prediction. can be used for the same future performance prediction task with roughly similar prediction results.

[0193]

[0194] in represents the predicted probability of the intervention sequence, and cat[·] represents the concatenation along the feature dimension. are trainable parameters that are not shared across operations.

[0195] Finally, the binary cross entropy loss function is used to measure the error between the predicted probability and the actual outcome. The loss for each intervention sequence is defined as follows:

[0196]

[0197] where r t+1 ∈{0,1} represents the actual response result at time t+1. For different intervention sequences, we calculate the loss Enables the model to extract stable and invariant causal relationships from different sequences.

[0198] (4-2) Problem Difficulty Prediction

[0199] like Figure 2 As shown in , the trivial subsequence opens a backdoor path through the causal chain C←S→T→H→Y. Therefore, the most effective way to cut off the path to future performance P is to link the trivial subsequence to another prediction target. In addition, for the intervention sequence, if the model converts the trivial subsequence Set to empty to intervene in the sequence will be the same as the original sequence. Then, Naturally, they can also be the same, allowing the model to easily achieve the prediction task through shortcuts. This goes against our original intention of changing the context of the causal subsequence.

[0200] Therefore, we need to introduce a new dynamic prediction target for trivial subsequences. Trivial subsequences contain response units that are irrelevant to future questions, which can be compared to a novice's first attempt at a new question. This response may tend to be towards the average accuracy of the group. In addition, since the accuracy is a discrete value, directly using it as a prediction target may lead to an imbalance between sample categories. Therefore, we discretize the continuous values ​​into difficulty levels and use the problem difficulty as the prediction target for trivial subsequences. Specifically, we first use all the answer records to calculate the average accuracy of the questions, then sort the accuracy and divide them into nd difficulty levels, each level containing the same number of questions. The calculation is as follows:

[0201]

[0202] where d t+1represents the difficulty level of the question to be answered at time t+1, f is the mapping function from accuracy to difficulty level, and δ(·) is the indicator function, which returns 1 if the condition is true, otherwise it returns 0. ij and r ij Represent a question and answer in all records respectively. Then, we use the trivial subsequence representation To predict the difficulty level of future questions, as follows:

[0203]

[0204] in Represents the predicted probability of the difficulty of the question. and are trainable parameters. cat[·] represents the concatenation operation along the feature dimension. Finally, we use binary cross entropy to measure the error between the predicted probability and the actual difficulty level of trivial subsequences:

[0205]

[0206] in is the loss value of the problem difficulty prediction task, I(d t+1 ) indicates the difficulty d t+1 Whether it belongs to the corresponding difficulty category. Unlike some existing practices, we do not use question difficulty as a feature or embedding input, but only as a target for predicting output. For example, if question difficulty is used for test question representation, the model may find shortcuts for difficulty prediction (for example, in the difficulty prediction formula, the model can t+1 Rather than ), the constraint on trivial subsequences is lost.

[0207] (4-3) Calculation of total loss

[0208] The overall loss is calculated by combining the losses of the future performance prediction task and the problem difficulty prediction task. The overall loss is defined as follows:

[0209]

[0210] where λ1, λ2, λ3, and λ4 are weighting coefficients used to balance the contribution of each loss term. correspond to the losses of deletion, substitution and reversal operations respectively, and is the loss of predicting trivial subsequences, i.e., the loss value of the problem difficulty prediction task.

[0211] The present invention experiments CIKT on three real-world public datasets and compares it with 10 baseline models. The evaluation indicators are the area under the curve (AUC), accuracy (ACC) and root mean square error (RMSE) commonly used in this field. The experimental results are shown in Table 1. CIKT achieves the best performance on all datasets, proving that this method can effectively improve the accuracy of future performance prediction.

[0212] Table 1 Comparison of prediction accuracy with 10 baseline models on three public datasets

[0213]

[0214] In addition, the robustness of the model is further verified. Specifically, different proportions of data are selected from the original ASSIST2012 dataset to create new training sets, and the test set is kept unchanged, and CIKT is compared with two baseline models (DKT is one of the most classic knowledge tracing models, and FKT is one of the most advanced knowledge tracing models). The experimental results are shown in Table 2. CIKT achieves the best performance in training sets of different sizes, especially when the training data is reduced to 1%, CIKT's advantage expands to 12.7% and 7.5%, which proves that CIKT has strong generalization ability and can work and achieve leading level even with very small amount of training data.

[0215] Table 2 Comparison results of prediction accuracy with two baseline models on the ASSIST2012 dataset

[0216]

[0217] The contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0218] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A knowledge tracking method based on causal invariance decoupling sequence, characterized in that: include: Firstly, a knowledge tracking model based on causal invariance decoupled sequence is constructed and trained. The knowledge tracking model includes a reaction sequence encoder, a causal identifier based on attention mechanism, a sequence causal intervener and a future performance predictor. The total loss of the model is obtained by multi-task learning method, and finally the trained model is used to realize knowledge tracking. The response sequence encoder is used to obtain the response sequence and the test question sequence, and the long short-term memory network is used as an encoder to encode the response sequence and the test question sequence to obtain the representation of the response sequence and the test question sequence; Among them, the learner at time t and a question q t The interaction of composes a reaction unit x t =(q t ,c t ,r t ), each unit contains question q t 、The knowledge concepts involved t and the reaction result r t , the reaction sequence is composed of multiple reaction units combined in the order of answering; The causal identifier based on the attention mechanism is used to construct a question-concept graph using the response sequence, and encode the question-concept graph through a graph convolutional network to obtain node representations of questions and concepts. The causal attention mechanism is used to identify the relationship between future questions and response sequences from the perspective of time and knowledge association, obtain the time association weight and knowledge association weight, and finally calculate the causal score and trivial score based on the time association weight and knowledge association weight; The sequence causal interventioner is used to decompose the original complete response sequence into two parts: a causal subsequence and a trivial subsequence by using the causal score and the trivial score. The intervention sequence is generated by performing a causal intervention operation on the trivial subsequence and combining it with the causal subsequence. Finally, the encoder and pooling operations are used to obtain the intervention sequence representation and the trivial subsequence representation. The future performance predictor is used to set the same future performance prediction task for different intervention sequence representations to discover the invariant and stable causal relationships therein. For trivial subsequence representations, a problem difficulty prediction task is introduced to eliminate prediction shortcuts, and the weights of the two tasks are balanced and optimized through multi-task learning to achieve the prediction of learners' future performance and the difficulty level of future problems.

2. According to claim 1, a knowledge tracking method based on causal invariance decoupling sequence is characterized by: The process of encoding a reaction sequence is as follows: Feature embedding: Use random embedding technology to map the elements in the reaction unit into a high-dimensional vector space to obtain an embedded representation of the reaction sequence. X T [x1,x2,…,x t ,…,x T ],x t =q t +c t +r t in represents the embedded representation of the reaction sequence, T represents the total length of the reaction sequence, and the embedded representations of the question, knowledge concept and reaction are respectively expressed as where d is the embedding dimension, represents the reaction unit representation at time t; Response sequence encoding: Use the long short-term memory network as an encoder to model the knowledge acquisition and forgetting process in students’ learning, receive the embedded representation of students’ historical response sequences and encode them to obtain the potential knowledge state representation; h t =z t ⊙tanh(s t ), in represents the potential knowledge state of the reaction unit at time t, and Represents the cell state, which transmits information between different moments; represents the output gate, represents the forget gate, represents the input gate, represents the candidate cell state, and thus the overall characterization of the reaction sequence is as follows: H T =LSTM X (X T )=[h1,h2,…,h t ,…,h T ] in Each row in represents the potential knowledge state of the corresponding reaction unit at each moment. The LSTM network for reaction sequence encoding is formally defined as LSTM X .

3. The knowledge tracking method based on causal invariance decoupling sequence according to claim 1 is characterized by: The process of encoding the test sequence in step (1) is as follows: The embedding representation of the test sequence is as follows: O T+1 =Q T+1 +C T+1 =[(q1+c1),(q2+c2),…,(q t +c t ),…,(q T+1 +c T+1 )] in represents the test question sequence embedding representation, and Represent the problem and concept sequence respectively; The embedded representation of the input test sequence is then encoded using the LSTM network as follows: E T+1 =LSTM E (O T+1 )=[e1,e2,…,e t+1 ,…,e T+1 ] in Each row in represents the representation of the test question to be answered. Corresponding to the question at time t+1, the LSTM used to encode the question sequence is denoted as LSTM E , whose structure is similar to LSTM X The same, but the parameters are not shared.

4. The knowledge tracking method based on causal invariance decoupling sequence according to claim 1 is characterized by: The problem concept graph constructed based on the reaction sequence in step (2) is represented as Node Set It consists of the questions and knowledge concepts involved in the reaction sequence. The same questions or concepts at different times are regarded as different nodes. First, the graph node representation is initialized. Specifically, it is defined first: The edge set ε is bidirectional, including sequence relationship edges and question-concept edges. The sequence relationship edges are added between the nodes of adjacent reactions, and the question-concept edges are added between the corresponding nodes through the correspondence between the questions and the knowledge concepts. For graph network updates, GCN is used to aggregate v n The neighbor nodes of the next layer are obtained, and the update formula is as follows: in represents the updated node representation of the nth node in the l+1th layer, in is the adjacency matrix A n The nth row in represents the node v n Relationships with other nodes; represents the embedding matrix of all nodes in layer l, represents the weight matrix of the lth layer, Dropout represents regularization, and ReLu is the activation function; after encoding the question-concept graph, the output representation of each node is expressed as Where L is the number of stacked GCN layers.

5. The knowledge tracking method based on causal invariance decoupling sequence according to claim 1 is characterized by: The specific implementation method of obtaining the time association weight and knowledge association weight in step (2) is as follows: First, from the perspective of time association, the test questions at time t+1 are represented as e t+1 Set as the query vector, set the reaction unit representation containing time information as the key vector, and use scaled dot product attention. The weight between the unit at t+1 and η is calculated as follows: in, represents the representation of the future test question at time t+1, represents the reaction unit representation at time η, θ t+1,η Represents the time-related weight; weight matrix and Linear transformations used in attention mechanisms, is the scaling factor; Secondly, from the perspective of knowledge association, we use the node representation after graph encoding to calculate the association between reaction units and represent the problem at time t+1 and conceptual representation Set it as the query vector, set the representation of other units in the sequence as the key vector, and the weight calculation is as follows: where ω t+1,η represents the knowledge association weight between the test questions at time t+1 in the future and other test questions at time η, ⊙ represents the Hadamard product, and multiplication is performed on the opposite elements; and is the problem and concept representation after graph encoding, the weight matrix and Used for linear transformations.

6. The knowledge tracking method based on causal invariance decoupling sequence according to claim 5 is characterized by: The specific implementation method of obtaining the causal score and trivial score in step (2) is as follows; The causal score is calculated as follows: in represents the causal score between the test question at the future time t+1 and the response unit at the time η, and They represent the time association weight and knowledge association weight used for causal score calculation respectively; is a masking matrix used to implement timing masking, with the upper triangular elements set to -∞ and the rest set to 0. t+1,μ is the masking value corresponding to t+1 and μ; The trivial score is calculated as follows: in represents a trivial fraction, and denote the time association weight and knowledge association weight used for trivial score calculation, respectively, and σ(·) denotes the sigmoid activation function; Furthermore, in order to completely decompose the reaction sequence into causal and trivial subsequences, the scores of continuous values ​​are converted to discrete values. This is achieved using the gumbel-softmax function as shown below: The gumbel-softmax function converts the continuous input vector into a discrete output vector while allowing the gradient to back propagate. and In {0,1}, therefore Finally, the discretized attention weights are again time-masked to ensure that no information leakage will occur in the future; specifically, for the test questions at the future time t+1, if η≥t+1, then and Set to 0, so the attention score of the test at time t+1 in the entire sequence is as follows: in and Represent the final causal score and trivial score of the entire sequence respectively; in addition, the sum of the causal score and trivial score matrices formed by all future test questions is a unit lower triangular matrix, in which all elements on the diagonal and below the diagonal are 1, and the other elements are 0, expressed as in 7. The knowledge tracking method based on causal invariance decoupling sequence according to claim 6 is characterized by: The causal score and trivial score are multiplied with the entire reaction sequence to extract the corresponding causal subsequence and trivial subsequence as follows: in and Denote the causal subsequence and trivial subsequence corresponding to time t+1, respectively. Each row in the matrix represents the embedding of a reaction unit. The non-causal or non-trivial part is filled with a zero vector through the Hadamard product. Thus, the reaction sequence is decoupled into two parts, and there is no overlap between the two subsequences, that is, Will After intervention t+1 The conditional probability is expressed as follows: in Represents Z t+1 The set of all possible values, do(·) represents the Do operator; This is approximated by considering a set of operations applied to trivial subsequences, resulting in a fixed number of intervening sequences, thus transforming it into the following approximation: where c t+1 Represents a trivial subsequence After the operation, the A collection,consists of causal subsequences with different contexts; Next, for the trivial subsequence Perform specific intervention operations to obtain the intervention sequence set c t+1 as follows: in represents an intervention sequence, oper(·) represents three different intervention operations, including removal (rem), replacement (repl), and inversion (inve); (3-3) Intervention sequence representation Finally, each intervention sequence is encoded to obtain the corresponding representation. The LSTM encoder is applied to each intervention sequence to capture the sequence dependency and encode the sequence into a fixed-length representation. The overall representation of the sequence is then obtained through an average pooling operation. The formula is as follows: in Represents intervention sequence The average pooling representation of t+1 represents a set of representations; in particular, It can also be seen as a causal subsequence Representation; LSTM X represents the response sequence encoder; The representation of a trivial subsequence is calculated as follows: in represents a trivial subsequence representation, and rem(·) represents a removal operation.

8. The knowledge tracking method based on causal invariance decoupling sequence according to claim 7 is characterized by: The future performance prediction tasks specifically include: Based on the causal subsequence, trivial subsequence and future test representation, the conditional probability of future performance is defined as follows: This shows that future performance t+1 is affected by the causal subsequence and the representation of future questions, while the trivial subsequence has no direct causal effect on prediction; therefore, different intervention sequences can be used for the same future performance prediction task, and their prediction results are roughly similar; in represents the predicted probability of the intervention sequence, and cat[·] represents the concatenation along the feature dimension. and are trainable parameters that are not shared across operations; Finally, the binary cross entropy loss function is used to measure the error between the predicted probability and the actual outcome. The loss for each intervention sequence is defined as follows: where r t+1 ∈{0,1} represents the actual response result at time t+1. For different intervention sequences, the loss is calculated separately. Enables the model to extract stable and invariant causal relationships from different sequences.

9. The knowledge tracking method based on causal invariance decoupling sequence according to claim 8 is characterized by: The problem difficulty prediction task specifically includes: Discretize continuous values ​​into difficulty levels and use question difficulty as the prediction target for trivial subsequences; specifically, first use all answer records to calculate the average correct rate of questions, then sort the correct rates and divide them into nd difficulty levels, each level contains the same number of questions, calculated as follows: where d t+1 represents the difficulty level of the question to be answered at time t+1, f is the mapping function from accuracy to difficulty level, δ(·) is the indicator function, which returns 1 if the condition is true, otherwise it returns 0; q ij and r ij Represent a question and answer in all records respectively; then, use trivial subsequence representation To predict the difficulty level of future questions, as follows: in represents the predicted probability of the difficulty of the question, and is a trainable parameter, cat[·] represents the cascade operation along the feature dimension; finally, binary cross entropy is used to measure the error between the predicted probability and the actual difficulty level of the trivial subsequence: in is the loss value of the problem difficulty prediction task, I(d t+1 ) indicates the difficulty d t+1 Whether it belongs to the corresponding difficulty category.

10. The knowledge tracking method based on causal invariance decoupling sequence according to claim 9 is characterized by: Combining the losses of the future performance prediction task and the problem difficulty prediction task, we calculate the overall loss, which is defined as follows: where λ1, λ2, λ3, and λ4 are weighting coefficients used to balance the contribution of each loss term, and correspond to the losses of deletion, substitution and reversal operations respectively, and is the predicted probability of the problem difficulty.

Citation Information

Patent Citations

  • Knowledge tracking method and system based on external attention mechanism

    CN114861754A

  • Knowledge tracking method based on graph attention network

    CN114911975A

  • Artificial intelligence tutoring system that support diagnosis of learning proficiency

    KR102538340B1

  • Intelligent tutoring system and method using knowledge tracing model based on transformer neural network

    KR102623059B1

Cited By

  • Task scene spatio-temporal data prediction method, system and device and medium

    CN120180073A