Knowledge tracking method and device based on anti-fact personalized enhancement

Abstract

The invention discloses a knowledge tracking method and device based on anti-fact personalized enhancement, and belongs to the technical field of knowledge tracking. The method comprises the steps of obtaining student information, questions and answer results, defining knowledge tracking, constructing a knowledge tracking causal graph based on the knowledge tracking, constructing an answer result prediction model based on the knowledge tracking causal graph, and predicting the answer results. The answer result prediction model comprises a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question characterization influence path sub-model and a fusion output sub-model; calculating a path total effect, a path direct effect and a path indirect effect based on a knowledge tracking causal graph; based on the path total effect, the path direct effect, the path indirect effect and the answer result prediction model, obtaining an anti-fact prediction depolarization formula of the question answer result of the student; based on an anti-fact prediction depolarization formula, an anti-fact prediction value of a question answering result of the student is obtained, and the method reduces the global deviation of the question in the knowledge tracking task.

Description

Technical Field

[0001] The present application belongs to the field of knowledge tracking technology, and in particular relates to a knowledge tracking method and device based on counterfactual personalized enhancement. Background Art

[0002] In the context of the deep integration of artificial intelligence and education, knowledge tracking, as a core technology for evaluating learning outcomes, and its modeling accuracy are crucial for the development of personalized education. The dynamic evolution of knowledge states directly influences the effectiveness of learning path planning. Systematic deviations in the modeling process can lead to inaccurate teaching decisions, which in turn affects the quality of talent development.

[0003] Existing knowledge tracking methods, primarily based on deep learning frameworks, still face significant technical bottlenecks. First, binary evaluation mechanisms struggle to capture the multidimensional characteristics of learning behavior, limiting the model's ability to represent complex cognitive states. Second, global bias issues trigger the capture of spurious correlated features, causing knowledge state assessment results to deviate from the true cognitive level. Furthermore, traditional bias correction methods lack effective mechanisms for maintaining the authenticity of data distribution, making it difficult to balance model robustness with practical application adaptability.

[0004] Related technologies simply use binary results to infer students' knowledge status, which can easily lead to over-reliance on the distribution of answers to these questions, and incorrectly interpret high or low accuracy as students' knowledge mastery level, thereby ignoring other factors (such as actual comprehension ability). There is a problem of failing to fully capture students' knowledge status that changes dynamically over time, and low accuracy in predicting students' knowledge status. Summary of the Invention

[0005] This application aims to solve at least one of the technical problems existing in the prior art. To this end, this application proposes a knowledge tracking method and device based on counterfactual personalized enhancement, which reduces the global bias of problems in knowledge tracking tasks.

[0006] In a first aspect, the present application provides a knowledge tracking method based on counterfactual personalized enhancement, the method comprising: Obtain student information, questions, and answers, and define knowledge tracking, which is to predict the probability of a student's answer to a question being correct based on the student's answers to historical questions and the student information; Constructing a knowledge tracking causal graph based on knowledge tracking, wherein the knowledge tracking causal graph includes a node set and a path set, wherein the node set includes students, questions, knowledge states, and answer results, and the path set includes a learning tendency driving path, a question representation influencing path, and a knowledge state mapping path; Constructing an answer result prediction model based on the knowledge tracing causal graph, the answer result prediction model includes a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question representation impact path sub-model and a fusion output sub-model; Calculate the total path effect, the direct path effect, and the indirect path effect based on the knowledge tracing causal graph; Based on the total path effect, the direct path effect, the indirect path effect and the answer result prediction model, a counterfactual prediction debiasing formula for the student's answer result to the question is obtained; Based on the counterfactual prediction debiasing formula, the counterfactual prediction value of the student's answer to the question is obtained.

[0007] According to one embodiment of the present application, the learning tendency driving path is student-answer result, which is used to describe the direct impact of student cognition on the answer result; the problem representation influence path is question-answer result, which is used to describe the direct impact of the question's own attributes on the answer result; and the knowledge state mapping path is student and question-knowledge state-answer result.

[0008] According to one embodiment of the present application, the knowledge state mapping path includes an ability-problem matching path and an answer result prediction path. The ability-problem matching path is student and problem-knowledge state, which is used to describe the decisive effect of the matching degree between student ability and problem on the knowledge state. The answer result prediction path is knowledge state-answer result, which is used to describe the predictive effect of knowledge state on answer result.

[0009] According to one embodiment of the present application, the learning tendency driven path sub-model includes a first fully connected network and a first multi-layer perceptron, which are used to obtain the learning tendency driven path prediction results, the knowledge state mapping path sub-model is a knowledge tracking model, the problem representation influence path sub-model includes a second fully connected network and a second multi-layer perceptron, which are used to obtain the problem representation influence path prediction results, and the fusion output sub-model includes a third multi-layer perceptron, which is used to obtain the predicted value of the student's answer to the question.

[0010] According to one embodiment of the present application, the calculation formula of the fusion output sub-model is as follows:

[0011] in, In fusion state, is the fusion function, is the intermediate output state of the learning tendency driving pathway sub-model, The knowledge mastery state of the knowledge state mapping path sub-model, The intermediate output state of the problem representation impacts the path sub-model, Indicates splicing, is a trainable matrix, is the trainable bias vector, is the predicted value of the answer to the question, It is the calculation value of the answer to the real-world question, and MLP is the third multi-layer perceptron.

[0012] According to one embodiment of the present application, the calculation formula of the counterfactual prediction debiasing formula is as follows:

[0013] in, is the predicted value of the answer to the question, Predicting outcomes for learning propensity-driven pathways, The problem characterization affects the path prediction results, TIE is the path indirect effect, NDE is the path direct effect, The weight of the predicted results of the learning tendency driving path, The weight that affects the path prediction result for the problem representation, is a hyperparameter, is the counterfactual prediction value of the student's answer to the question, S is the student, Q is the question, and Y is the student's answer result. is the local learning tendency, t is the time step, is the global question representation of the question answered at time step t+1.

[0014] According to one embodiment of the present application, the training process of the answer result prediction model includes: Construct a prediction model for preset answer results; Obtain different student information, different questions, and different answer results as a data set; Based on the loss function, a preset answer result prediction model is trained according to the data set to obtain the answer result prediction model, wherein the loss function is constructed based on the binary cross entropy loss function.

[0015] In a second aspect, the present application provides a knowledge tracking device based on counterfactual personalized enhancement, the device comprising: An acquisition module is used to obtain student information, questions, and answer results, and define knowledge tracking, which is to predict the probability of a student's answer to a question being correct based on the student's answer results to historical questions and student information; A first construction module is configured to construct a knowledge tracing causal graph based on knowledge tracing, wherein the knowledge tracing causal graph includes a node set and a path set, wherein the node set includes students, questions, knowledge states, and answer results, and the path set includes a learning tendency driving path, a question representation influencing path, and a knowledge state mapping path; A second construction module is configured to construct an answer result prediction model based on the knowledge tracing causal graph, wherein the answer result prediction model includes a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question representation impact path sub-model, and a fusion output sub-model; A first calculation module is used to calculate the total effect of the path, the direct effect of the path and the indirect effect of the path based on the knowledge tracing causal graph; A second calculation module is used to obtain a counterfactual prediction debiasing formula for the student's answer to the question based on the total path effect, the direct path effect, the indirect path effect and the answer result prediction model; The prediction module is used to obtain the counterfactual prediction value of the student's answer to the question based on the counterfactual prediction debiasing formula.

[0016] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the knowledge tracking method based on counterfactual personalized enhancement as described in the first aspect above is implemented.

[0017] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the knowledge tracking method based on counterfactual personalized enhancement as described in the first aspect above.

[0018] In a fifth aspect, the present application provides a chip comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the knowledge tracking method based on counterfactual personalized enhancement as described in the first aspect.

[0019] In a sixth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the knowledge tracking method based on counterfactual personalized enhancement as described in the first aspect above.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become obvious from the description below, or will be learned through practice of the present application.

[0021] The present invention provides a knowledge tracking method based on counterfactual personalized enhancement, which has the following advantages over the prior art: (1) The present invention obtains student information, questions, and answer results, and defines the knowledge tracing problem based on this. In combination with the counterfactual prediction debiasing formula, the present invention calculates the total path effect, direct path effect, and indirect path effect of the knowledge tracing causal graph, and comprehensively utilizes deep learning and causal inference technology to explore students' local learning tendencies and global problem representations. A method that can support different knowledge tracing models is designed, which captures students' local learning tendencies and global problem representations, and effectively alleviates the global bias in the knowledge tracing task by using counterfactual inference, thereby enhancing students' personalized performance.

[0022] (2) The present invention calculates the answer to the question by adopting the counterfactual prediction debiasing formula, combines the prediction results of the learning tendency driving path and the problem representation influencing path, adjusts the direct effects of the learning tendency driving and problem representation influencing paths, and combines the indirect effects of the knowledge state mapping path to achieve effect-controlled counterfactual prediction, which can alleviate the global bias in the knowledge tracking task, improve the students' prediction accuracy of the answer to the question, help to more accurately analyze the students' knowledge mastery, and provide a more reliable basis for the formulation of personalized learning strategies.

[0023] (3) The present invention improves the accuracy of the answer result prediction model by fusing the intermediate output states of the learning tendency driving path sub-model, the knowledge state mapping path sub-model and the problem representation influencing path sub-model through the fusion output sub-model, combining the trainable matrix and the bias vector, and using the multi-layer perceptron to make the final prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1 1 is a flow chart of a knowledge tracking method based on counterfactual personalized enhancement provided by an embodiment of the present application; Figure 2 A schematic diagram of the structure of a knowledge tracing causal graph provided in an embodiment of the present application; Figure 3 This is a diagram of the knowledge tracking framework structure provided by an embodiment of the present application; Figure 4 Schematic diagram of the structure of the knowledge tracking device based on counterfactual personalized enhancement provided by an embodiment of the present application; Figure 5 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the accompanying drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of this application.

[0026] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0027] Below, in conjunction with the accompanying drawings, the knowledge tracking method based on counterfactual personalized enhancement, the knowledge tracking device based on counterfactual personalized enhancement, the electronic device and the readable storage medium provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios.

[0028] Among them, the knowledge tracking method based on counterfactual personalized enhancement can be applied to a terminal, and can be specifically executed by hardware or software in the terminal.

[0029] The terminal includes, but is not limited to, a portable communication device such as a mobile phone or tablet computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad). It should also be understood that in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad).

[0030] In the following embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse, and a joystick.

[0031] The embodiment of the present application provides a knowledge tracking method based on counterfactual personalized enhancement. The execution subject of the knowledge tracking method based on counterfactual personalized enhancement can be an electronic device or a functional module or functional entity in the electronic device that can implement the knowledge tracking method based on counterfactual personalized enhancement. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablets, computers, cameras and wearable devices, etc. The knowledge tracking method based on counterfactual personalized enhancement provided in the embodiment of the present application is explained below using electronic devices as the execution subject as an example.

[0032] With the rapid development of artificial intelligence, integrating advanced technologies with education has become a key trend in recent years. Students are no longer satisfied with traditional face-to-face learning and are increasingly turning to online platforms and intelligent educational tools. The continuous accumulation of data on online learning activities provides valuable opportunities to understand students' learning processes and better meet personalized educational needs.

[0033] During the learning process, students develop an internal knowledge state that reflects their mastery of the content and is continuously updated as learning progresses. To understand the learning process behind students, tracking the evolution of their knowledge state is crucial. This is the crux of knowledge tracing technology. Knowledge tracing aims to model student knowledge mastery based on learning records and predict future performance.

[0034] Knowledge tracing tasks typically assume binary answers, meaning each question has only two possible outcomes: "correct" or "incorrect." However, this simplified assessment method may not reflect students' mastery, especially when students exhibit biases on certain questions. For example, some overly simple questions may result in a high percentage of students answering correctly, while some more challenging questions, due to their difficulty, may produce more incorrect answers. If a model simply uses these binary results to infer students' knowledge, it can easily become overly reliant on the distribution of responses to these questions. Specifically, the model can take a shortcut, mistakenly interpreting high accuracy as indicative of students' knowledge mastery, thereby ignoring other factors (such as actual comprehension) and incorrectly concluding that students have mastered a particular knowledge point, even though they may not have performed well on these complex questions. Consequently, this global bias, caused by the imbalance in the proportion of correct and incorrect answers, not only causes the model to overfit these features but also fails to fully capture the dynamic changes in students' knowledge over time, leading to incorrect modeling of knowledge states.

[0035] To mitigate global bias, a simple solution is to resample the dataset to balance the proportion of correct and incorrect answers for each question. This addresses the root cause of bias, but in reality, this approach has certain limitations. First, it may alter the actual data distribution, affecting the model's applicability in real-world scenarios. Second, in real-world learning scenarios, some questions inherently carry significant bias due to their content. Therefore, simple resampling fails to effectively identify and eliminate the source of global bias.

[0036] Causal inference provides a systematic approach that can clarify the causal relationship between various factors by constructing a causal graph. From the perspective of causal inference, global bias can be regarded as the interference of confounding variables on model predictions. Therefore, by introducing counterfactual inference, under the premise of controlling the influence of confounding variables, it is possible to estimate the student's performance in answering when not affected by the bias. On the other hand, if we carefully analyze the global bias in knowledge tracking, we can find that global bias affects model predictions from a global perspective, that is, it affects the model's inference of the student's knowledge status based on the global problem representation. Therefore, another key point to solve this problem is to introduce a more sophisticated personalized modeling strategy, taking local learning tendencies into consideration, and avoiding the model from relying solely on the global data distribution. By defining the student's local learning tendencies, the student's strengths and weaknesses are revealed, helping the model identify learning needs that are not fully reflected by the global bias and make inferences that are more in line with the student's learning status.

[0037] Although causal inference has been widely used in various fields, research on integrating it into knowledge tracing tasks is still in its early stages. Therefore, exploring a knowledge tracing method based on counterfactual personalized enhancement, which uses counterfactual inference, personalized modeling, and deep learning techniques to model the evolution of student knowledge, has important research significance and application value for mitigating bias and improving the inference of knowledge states.

[0038] Figure 1 : is a flow chart of the knowledge tracking method based on counterfactual personalized enhancement provided by the embodiment of the present application, such as Figure 1 As shown, the knowledge tracking method based on counterfactual personalized enhancement includes: step 110, step 120, step 130 and step 140.

[0039] Step 110: Obtain student information, questions, and answer results, and define knowledge tracking, wherein the knowledge tracking is to predict the probability of the student's answer to the question being correct based on the student's answer results to the historical questions and the student information; It is easy to understand that the knowledge tracking task dynamically models the changing process of the learner's knowledge state by analyzing the learner's historical learning sequence and predicting the future mastery of specific knowledge points.

[0040] In the knowledge tracking task, we first obtain student information, questions, and answer results, and define knowledge tracking, global problem representation, and local learning tendency. Student information includes student ID.

[0041] Specifically, in online learning platforms, students will leave behind massive amounts of historical data through learning exercises. Each student will answer a series of questions related to specific knowledge concepts and give corresponding answers, thereby promoting knowledge acquisition and mastery.

[0042] For example, assume that the student set has I different students, represented as follows:

[0043] in, For the i-th student.

[0044] The problem set has N different problems, which are represented as follows:

[0045] in, This is the nth question.

[0046] The set of knowledge concepts is the set of knowledge concepts involved in all problems. A knowledge concept is represented as follows:

[0047] in, This is the mth question.

[0048] Students will answer questions containing knowledge concepts at time step t and get an answer result, which is expressed as:

[0049] in, is the answer result of the question at time step t, 1 means the answer is correct and 0 means the answer is wrong.

[0050] Therefore, each student's learning sequence can be viewed as a record of interactions with the exercises over multiple time steps, represented as follows:

[0051] in, For a single question, The knowledge concepts involved in the problem, The answer result of a single question.

[0052] It is easy to understand that the student's learning sequence contains the evolution of the knowledge state. The knowledge tracking task aims to evaluate the student's knowledge state through the historical learning sequence and then predict whether they can correctly answer the next question. The calculation formula is as follows:

[0053] It should be noted that knowledge tracking mainly relies on the interaction records between students and problems to obtain knowledge status, but these interaction records lack student information and ignore the differences in individual students' ability to absorb knowledge. Because even with the same interaction records, due to differences in each student's cognitive ability, learning strategy, learning motivation and other factors, they will show different knowledge state growth for the same learning content. Therefore, simply using interaction records to model knowledge status may not reflect the actual situation, and incorporating student information and interaction records into model construction can more accurately portray students' knowledge status. Therefore, the calculation formula for the final knowledge tracking task is as follows:

[0054] Global problem representation uses problem difficulty to reflect the degree of global bias. Global bias usually stems from the attributes of the problem itself and the answer pattern of highly exposed problems in the data, which makes the correct and incorrect answers to the questions unevenly distributed. If this phenomenon is expressed statistically, it can usually be expressed by problem difficulty. Therefore, global problem representation, that is, problem difficulty, is used here to reflect the degree of global bias. Since problems are related to knowledge concepts, the same problem involves a certain knowledge point, but the same knowledge point involves different problems. The difficulty brought by the knowledge point will be different from the difficulty of the problem itself. In addition, the degree of exposure also affects the reliability of the difficulty assessment. If the exposure rate is low, the difficulty measured by the correct rate of the question will be unreliable. Therefore, the difficulty of the global problem is jointly adjusted using the difficulty of the question and the difficulty of the knowledge point. According to Bayesian estimation, the difficulty of the knowledge concept is first calculated. The calculation formula is as follows:

[0055] in, is the number of times the knowledge concept is answered, is the correct answer rate of knowledge concepts, is the average number of answers to all knowledge concepts, is the average correct answer rate for all knowledge concepts, For the concept of knowledge, The difficulty of knowledge concepts.

[0056] Then, the difficulty of the knowledge concept is used to adjust the difficulty of the question. The formula for calculating the difficulty of the question is as follows:

[0057] in, is the number of times the question was answered, is the correct answer rate of the question, is the average number of times all questions are answered, is the difficulty of the knowledge concept corresponding to the question, is the difficulty of the j-th question.

[0058] The local learning tendency of students is to dynamically reflect their personalized learning status and progress trend by analyzing their answering performance at time steps. The calculation of local learning tendency is based on the student's answer results and the global problem representation. Calculate the difference between the student's answer to the question and the global problem representation, and then calculate the cumulative average of the difference over historical time steps, which is the student's local learning tendency at different time steps. A student's local learning tendency can help alleviate global bias. Even if a question appears to be relatively difficult in the dataset, it does not necessarily present the same challenge to all students. For some students, the question may still be answered correctly. This requires considering individual differences by considering the student's learning tendency.

[0059] The local learning tendency can be obtained by combining the student's answer results with the global problem representation. The student's answer results to each question can reflect his or her learning performance, but due to the different difficulty levels of the questions, the value reflected by correct or incorrect answers is not exactly the same. If a more difficult question is answered correctly, it may be more indicative of the student's excellent performance. Conversely, if a less difficult question is answered incorrectly, it may indicate that the student still has weaknesses. Local learning tendencies need to reflect the student's personalized performance more, and the answer results themselves contain information about the difficulty of the question, so the influence of the question itself on the answer results needs to be eliminated. In addition, local learning tendencies need to dynamically characterize the student's performance at different time steps, so it is necessary to calculate their cumulative performance from the beginning of learning to the current time step. Specifically, for the time step For the above question, we first need to calculate the difference between the answer and the global question representation to correct the answer performance to make it more consistent with the individual learning situation. Then, we calculate the cumulative mean of this difference over historical time steps to make it reflect the dynamic learning trend and avoid the random influence of a single answer. The specific formula is as follows:

[0060] in, is the time step The answer result when is the time step The global problem representation of the question answered when is the local learning tendency and t is the time step.

[0061] Local learning tendency is somewhat interpretable. Correct answers to difficult questions will result in higher positive feedback, while incorrect answers to less difficult questions will have a greater negative impact. Local learning tendency allows for a more equitable assessment of a student's individual learning status and progress.

[0062] Step 120: construct a knowledge tracing causal graph based on knowledge tracing, wherein the knowledge tracing causal graph includes a node set and a path set, wherein the node set includes students, questions, knowledge states, and answer results, and the path set includes a learning tendency driving path, a question representation influencing path, and a knowledge state mapping path; Figure 2 A schematic diagram of the structure of the knowledge tracing causal graph provided in the embodiment of the present application is shown as follows: Figure 2 As shown, the knowledge tracing causal graph includes a node set and a path set. The node set includes students, questions, knowledge status and answer results. Among them, node S represents students, node Q represents questions, node H represents student knowledge status, and node Y represents student answer results.

[0063] It should be noted that most knowledge tracking models are based on Figure 2 The causal diagram in (a) is constructed, but Figure 2 The causal diagram in (b) more accurately depicts the model's prediction process. Specifically, the student and the question itself directly influence the prediction results. The former encompasses individual differences among students, and their inherent characteristics can have a direct impact on the answer. For example, a student with strong learning ability is likely to answer correctly even when faced with a difficult question. It is important to note that the difference between a student's learning ability and their knowledge state is that the knowledge state is local and only reflects the student's mastery of certain knowledge points; while learning ability is diverse, such as cognitive ability, which goes beyond the scope of the knowledge state. The latter incorporates the attribute characteristics of the question, as questions that are too simple or too difficult may directly lead to a student's correct or incorrect answer without relying too much on the student's knowledge state.

[0064] In some embodiments, using Figure 2 In the causal diagram (b), the path set includes the learning tendency driving path SY, the problem representation influencing path QY, and the knowledge state mapping paths S and QHY. The knowledge state mapping path can be decomposed into the ability-problem matching paths S and QH and the answer result prediction path HY.

[0065] Step 130: construct an answer result prediction model based on the knowledge tracing causal graph, wherein the answer result prediction model includes a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question representation impact path sub-model, and a fusion output sub-model; Furthermore, the answer result prediction model is constructed based on the knowledge tracing causal graph, which specifically includes the following steps: (1) Embedding representation: The input elements of the answer result prediction model are student information, questions, answer results, global question representation and local learning tendency. The global question representation and local learning tendency are represented by continuous embedding, and the other elements are embedded using embedding matrices. The learning sequence is regarded as the splicing of question / knowledge concepts and answer response embeddings.

[0066] (2) Constructing a learning tendency driving path sub-model: used to describe the learning tendency driving path in the causal graph, embedding the local learning tendency at time step t into the module, and using two fully connected networks to obtain the local learning tendency state, and finally using a multi-layer perceptron to obtain the prediction result.

[0067] (3) Constructing the knowledge state mapping path sub-model: It is used to describe the knowledge state mapping path in the causal graph. The existing knowledge tracking model generally follows this path to predict the answer result. The knowledge state mapping path sub-model is constructed based on the existing knowledge tracking model, and the input student embedding and interaction sequence embedding , obtain the knowledge state through the state update layer, and then obtain the output result through the mapping layer.

[0068] (4) Constructing a problem representation impact path sub-model: This sub-model is used to describe the problem representation impact path in the causal graph. The global problem representation at time step t+1 is embedded into this module, and the problem representation state is obtained using two fully connected networks. Finally, a multi-layer perceptron is used to obtain the prediction result.

[0069] (5) Constructing a fusion output sub-model: The answer result is jointly affected by the learning tendency driving path, the knowledge state mapping path, and the problem representation influencing path. Therefore, the state outputs of the learning tendency driving path sub-model, the knowledge state mapping path sub-model, and the problem representation influencing path sub-model are first spliced and fused to obtain a new knowledge mastery state, and then a multi-layer perceptron is used to obtain the predicted value of the answer result.

[0070] Step 140: Calculate the total path effect, the direct path effect, and the indirect path effect based on the knowledge tracing causal graph; The total path effect is the combined effect of the student and the question on the answer. Comparing the answer to a question in the real world with the answer in a counterfactual world where the variable is in a reference state (a state without a specific intervention or influence) is the total path effect. It measures the change in the answer between the two states.

[0071] The direct effects in the causal diagram encompass the effects of students and questions, respectively, on the answer through direct pathways. When the mediating variable, student knowledge status, is blocked, the change in the dependent variable directly caused by changes in the independent variable—the natural direct effect—can be calculated. By jointly analyzing the learning tendency-driven pathway and the problem representation-influencing pathway, their combined direct effects on the answer can be derived.

[0072] The knowledge state mapping path contains the mediating variable knowledge state. Students and questions indirectly influence the answer results through the mediating variable knowledge state, that is, the indirect effect of students and questions on the answer results. The total effect can be decomposed into a natural direct effect and a total indirect effect.

[0073] Step 150: Based on the total path effect, the direct path effect, the indirect path effect, and the answer result prediction model, a counterfactual prediction debiasing formula for the student's answer result to the question is obtained; It is worth noting that students' local learning tendencies help to reveal the impact of individual differences on the learning process, thereby alleviating global bias. Therefore, on the basis of indirect effects, by adjusting the direct effects of the paths, a debiasing formula that incorporates students' local learning tendencies is obtained. Based on this formula, combined with the implementation methods of each path, a debiasing formula for counterfactual predictions of students' answers to questions is obtained.

[0074] Step 160: Based on the counterfactual prediction debiasing formula, obtain the counterfactual prediction value of the student's answer to the question.

[0075] According to the knowledge tracking method based on counterfactual personalized enhancement provided in the embodiment of the present application, by obtaining student information, questions and answer results, and defining the knowledge tracking problem based on this, combined with the counterfactual prediction debiasing formula, by calculating the total path effect, path direct effect and path indirect effect of the knowledge tracking causal graph, deep learning and causal inference technology are comprehensively utilized to explore students' local learning tendencies and global problem representations, and a method that can support different knowledge tracking models is designed to capture students' local learning tendencies and global problem representations, and use counterfactual inference to effectively alleviate the global bias in the knowledge tracking task, thereby enhancing students' personalized performance.

[0076] In some embodiments, the learning tendency driving path is student-answer result, which is used to describe the direct impact of student cognition on the answer result; the problem representation influence path is question-answer result, which is used to describe the direct impact of the question's own attributes on the answer result; and the knowledge state mapping path is student and question-knowledge state-answer result.

[0077] It is easy to understand that the learning tendency-driven path has a direct impact on students' answer results. High-ability students can flexibly apply the knowledge points they have learned to improve the accuracy of their answers to questions, while low-ability students may be affected by their lack of strategies and their answer performance.

[0078] The question representation influencing path is the direct impact of the question's own attributes on the answer result. Questions that are too simple or too difficult may directly lead to students answering correctly or incorrectly, resulting in global bias, so that the model can make better predictions about the answer results only by remembering the characteristics of the question.

[0079] In this example, by inputting student information, questions, answer results, global question representations, and local learning tendencies into the learning tendency driving path, the question representation influencing path, and the knowledge state mapping path, a prediction of the student's answer to the next question is obtained. This enables a comprehensive analysis of the student's individual cognitive level, question characteristics, and knowledge state, improving the accuracy of predicted student answer results, providing strong support for personalized education and question design, and optimizing teaching effectiveness and learning experience.

[0080] In some embodiments, the knowledge state mapping path includes an ability-problem matching path and an answer result prediction path. The ability-problem matching path is student and problem-knowledge state, which is used to describe the decisive effect of the matching degree between student ability and problem on the knowledge state. The answer result prediction path is knowledge state-answer result, which is used to describe the predictive effect of knowledge state on answer result.

[0081] In some embodiments, the ability-question matching pathway determines the knowledge state of the student and the question. The higher the degree of match between the student's ability and the question, the higher the student's mastery of the question. The answer prediction pathway predicts the answer result based on the knowledge state.

[0082] In this embodiment, by combining the ability question matching path and the answer result prediction path to form a knowledge state mapping path, the student's knowledge mastery can be evaluated more comprehensively and accurately, thereby enhancing the student's personalized performance.

[0083] In some embodiments, the learning tendency driven path sub-model includes a first fully connected network and a first multi-layer perceptron, which are used to obtain the learning tendency driven path prediction results, the knowledge state mapping path sub-model is a knowledge tracking model, the problem representation influence path sub-model includes a second fully connected network and a second multi-layer perceptron, which are used to obtain the problem representation influence path prediction results, and the fusion output sub-model includes a third multi-layer perceptron, which is used to obtain the predicted value of the student's answer to the question.

[0084] Figure 3 This is a diagram of the knowledge tracking framework provided by the embodiment of the present application, such as Figure 3 As shown in the figure, the elements are first embedded and represented. Then, the answer result prediction model is constructed according to the path in the causal graph, including the learning tendency driving path sub-model, the knowledge state mapping path sub-model, the question representation influence path sub-model and the fusion output sub-model. Finally, the intermediate output states of the learning tendency driving path sub-model, the knowledge state mapping path sub-model and the question representation influence path sub-model are used to obtain the predicted value of the question answer result through the fusion output sub-model. The specific process is as follows: (1) Embedded representation The input elements of the framework are student information, questions, answer results, global problem representation and local learning tendency, and each input element is embedded. Specifically, the student embedding is achieved through the embedding matrix (in is the embedding dimension), the problem is embedded through the embedding matrix (in is the embedding dimension), knowledge concept embedding is represented by the embedding matrix (in is the embedding dimension), the answer response is embedded through the embedding matrix (in is the embedding dimension). The global problem representation embedding and the local learning tendency embedding are represented by continuous embedding, specifically and ,in and is a trainable vector.

[0085] Given a learning sequence, each interaction can be viewed as a concatenation of questions / knowledge concepts and answer responses, i.e. The choice of questions and knowledge concepts depends on the requirements of the knowledge tracking model.

[0086] (2) Learning tendency-driven path sub-model It is easy to understand that the model is constructed based on the learning tendency driving path and is used to model students' personalized local learning tendencies. The prediction result of the learning tendency driving path is defined as ,in Structural equations for modeling the pathways driving learning dispositions.

[0087] To more accurately model the personalized dynamic learning process, the time step Locally learned tendency embedding on express Specifically, we first use a simple network layer to capture the student's local learning tendency. This process uses a linear layer, uses the ReLU activation function, and then outputs it through another linear layer to obtain the local learning tendency state. The calculation formula for the local learning tendency state is as follows: , in, is the local learning tendency state, and is a trainable matrix, and is the trainable bias vector.

[0088] Then, a multi-layer perceptron (MLP) is used to perform nonlinear mapping on the local learning tendency state, and the prediction result output is: .

[0089] (3) Knowledge state mapping path sub-model It is easy to understand that this model corresponds to a knowledge state mapping path, and the output of this path is ,in It is a structural equation about the knowledge state update, It is a structural equation about the output. Existing knowledge tracking models generally follow this path for prediction, so this path is built based on existing knowledge tracking models. For example, for the deep knowledge tracking model DKT, using student embedding express , using the learning interaction sequence express , and use both as input to the state update module (RNN, LSTM) to obtain the updated knowledge state , and then get the output result through the mapping layer.

[0090] (4) Problem representation impact path sub-model The problem characterization is defined as the impact path prediction result: ,in The structural equations that characterize the influence paths for modeling problems. Similar to the learning tendency driven path, this module divides the time steps Global question representation embedding on express Specifically, we first use a linear layer and activation function to summarize the problem representation state. The calculation formula is as follows:

[0091] in, is the problem representation state, and is a trainable matrix, and is the trainable bias vector.

[0092] Then, a multi-layer perceptron (MLP) is used to obtain the prediction output. The calculation formula is as follows: .

[0093] In this embodiment, by fusing the intermediate output states of the learning tendency driven path sub-model, the knowledge state mapping path sub-model and the problem representation impact path sub-model, and using a multi-layer perceptron for final prediction, the output information of different path models is combined to improve the understanding of the student's learning status, thereby achieving more accurate personalized learning predictions, optimizing educational decision support, and further improving the accuracy of learning effect evaluation.

[0094] In some embodiments, the calculation formula of the fusion output sub-model is as follows:

[0095] in, In fusion state, is the fusion function, is the intermediate output state of the learning tendency driving pathway sub-model, The knowledge mastery state of the knowledge state mapping path sub-model, The intermediate output state of the problem representation impacts the path sub-model, Indicates splicing, is a trainable matrix, is the trainable bias vector, is the predicted value of the answer to the question, It is the calculation value of the answer to the real-world question, and MLP is the third multi-layer perceptron.

[0096] It is easy to understand that from the perspective of causality, Figure 2 The answer result in (b) is jointly affected by the learning tendency driving path, the knowledge state mapping path, and the problem representation influencing path. The state outputs of the learning tendency driving path module, the knowledge state mapping path module, and the problem representation influencing path module are 、 、 The goal of the three-path fusion is to predict the probability of answering the question correctly at the next time step, so it is necessary to use the time step on the three paths Time , used to predict the time step When answering the result and time steps The above question , and fuse them. According to the ancestor node, define the time step superior The calculation formula is as follows:

[0097] in, is the structural equation, is the fusion function.

[0098] It is worth noting that in order to ensure the clarity of the symbol expression, the uniform use of and Represents related variables in different paths, although they are represented differently in different paths. There are many ways to construct the fusion function. Here, the three states are concatenated and output: the predicted value is output through a multi-layer perceptron.

[0099] In some embodiments, the first multi-layer perceptron, the second multi-layer perceptron, and the third multi-layer perceptron are the same multi-layer perceptron.

[0100] In this embodiment, the fusion output sub-model integrates the intermediate output states of the learning tendency driving path sub-model, the knowledge state mapping path sub-model and the problem representation influencing path sub-model, combines the trainable matrix and the bias vector, and uses a multi-layer perceptron to make the final prediction. By fusing the output information of different paths, the accuracy of the answer result prediction model is improved.

[0101] In some embodiments, the calculation formula of the counterfactual prediction debiasing formula is as follows:

[0102] in, is the predicted value of the answer to the question, Predicting outcomes for learning propensity-driven pathways, The problem characterization affects the path prediction results, TIE is the path indirect effect, NDE is the path direct effect, The weight of the predicted results of the learning tendency driving path, The weight that affects the path prediction result for the problem representation, is a hyperparameter, is the counterfactual prediction value of the student's answer to the question, S is the student, Q is the question, and Y is the student's answer result. is the local learning tendency, t is the time step, is the global question representation of the question answered at time step t+1.

[0103] In some embodiments, expressions for the total path effect, direct effect, and indirect effect are obtained based on causal graph analysis. Then, by controlling the direct effect and combining the fusion output sub-model, a counterfactual prediction debiasing formula for answering the question is obtained. The specific process is as follows: (1) Calculate the total path effect The total effect is the change in the dependent variable when the independent variable moves from a reference state to a target state in causal inference. In this causal diagram, the total effect of the path is and right The total effect is the real world Results The variables in the counterfactual world are in the reference state Results The comprehensive reflection of the differences, the calculation formula of the total effect is as follows,

[0104] Where TE represents the total path effect, Computational values for answering real-world questions, The calculated value of the answer to the counterfactual world question.

[0105] It should be noted that , and It can be understood as the reference value of a variable, used to represent a hypothetical state. The total effect TE can be understood as the difference between the expected value and the reference value, that is, the total impact of the independent variable on the dependent variable.

[0106] (2) Calculation of direct effects exist Figure 2 In the causal diagram shown in (b), the direct effect is the effect of the student and the problem affecting the answer result through the learning tendency driving path and the problem representation influencing path respectively, which is called the natural direct effect. The natural direct effect represents the influence of the independent variable on the dependent variable through a direct path. Specifically, this effect reflects the change in the dependent variable caused by the change of the independent variable when the mediating variable is blocked. Figure 2 As shown in (c), when the mediating variable knowledge state Fixed to and The direct effect is calculated as follows:

[0107] Among them, NDE is the direct effect, The calculated value of the answer to the counterfactual world question without considering the knowledge state.

[0108] The direct effect estimates the effect that would occur if only students' learning dispositions and problem representations acted directly on the answer outcome in the counterfactual world.

[0109] (3) Calculation of indirect effects According to the causal graph structure, the total effect can be decomposed into natural direct effect and total indirect effect. The total indirect effect represents the impact on the dependent variable through the mediating variable. Figure 2In (b), the total effect of the student and question on the answer can be decomposed into the direct effects of the student and question on the answer via the direct paths of student-answer and question-answer, plus the indirect effects of the student and question on the answer via the mediating variable knowledge state via the knowledge state mapping path. The formula for calculating the total effect, direct effect, and indirect effect is as follows:

[0110] Among them, TIE is the indirect effect.

[0111] To mitigate global bias in knowledge tracking, the ultimate goal is to achieve bias mitigation by changing the causal effects of the two direct paths: student-answer result and question-answer result. First, we need to obtain the causal effects other than these two paths, namely the indirect effects. The formula for calculating the indirect effect of the knowledge state mapping path is as follows:

[0112] (4) Calculate the debiasing formula for the counterfactual prediction of the answer question The global bias of knowledge tracking can be considered the influence of the question-answer result path, and the student-answer result can be considered the effect of the student's local learning tendency. Global bias ignores the individual behavioral characteristics of individual students due to over-reliance on the general behavioral characteristics of all students. Local learning tendency focuses on the student's personalized understanding of the question, rather than relying solely on the overall data distribution. Using students' local learning tendencies helps to reveal the impact of individual differences on the learning process, thereby reducing the impact of global bias. Therefore, it is necessary to control the direct effect and retain the positive signal belonging to the student's local learning tendency. The initial calculation formula of the counterfactual prediction based on the debiasing formula is as follows:

[0113] in, is a hyperparameter used to control the impact of NDE on the prediction of future problem performance.

[0114] Furthermore, combined with the fusion output formula, the time step on It can be expressed as:

[0115] It is worth noting that since the knowledge state variable is in the reference state, it can be regarded as a constant term, and its impact on the prediction is consistent and can be directly ignored. The above formula can be simplified to:

[0116] for ,in 、 、 All three variables are in the reference state, that is, they are fixed and not affected by other changes. 、 and The causal effect of is assumed to be constant, representing a baseline situation. In this setting, It is regarded as a constant term and can be ignored.

[0117] Finally, the calculation formula of the counterfactual prediction based on the debiasing formula is as follows:

[0118] It should be noted that and To achieve better bias mitigation, we use adjustable hyperparameters to adjust the influence of students’ local learning tendencies and global problem representations.

[0119] (In this embodiment, by using a counterfactual prediction debiasing formula to calculate the question answer results, combining the prediction results of the learning tendency driving path and the problem representation influencing path, and by adjusting the direct effects of the learning tendency driving path and the problem representation influencing path, combined with the indirect effect of the knowledge state mapping path, effect-controlled counterfactual prediction is achieved. This can alleviate the global bias in the knowledge tracking task, improve the accuracy of students' predictions of question answer results, help more accurately analyze students' knowledge mastery, and provide a more reliable basis for the formulation of personalized learning strategies.

[0120] In some embodiments, the training process of the answer result prediction model includes: Construct a prediction model for preset answer results; Obtain different student information, different questions, and different answer results as a data set; Based on the loss function, a preset answer result prediction model is trained according to the data set to obtain the answer result prediction model, wherein the loss function is constructed based on the binary cross entropy loss function.

[0121] In some embodiments, the answer prediction model is trained using a multi-task learning approach. To evaluate the model's predictive performance, appropriate public knowledge tracking datasets are selected for training and prediction. These public datasets, such as ASSISTments2009 and ASSISTments2017, contain basic information such as student ID, knowledge concept ID, question ID, and answer results. The dataset is divided into a training-validation set and a test set in an 8:2 ratio, and then the training-validation set is further divided into a training set and a validation set in an 8:2 ratio. Traditional evaluation strategies typically make predictions based on randomly selected test data, but because these data still exhibit global bias, they cannot accurately reflect the model's predictive ability after causal inference. To test the model's debiasing ability, the test set is debiased by stratified sampling based on question category. Specifically, the correct and incorrect interactions for each question category in the test set are sampled so that the number of correct and incorrect answers is equal. To ensure the size of the entire test set remains constant, some question interactions may be duplicated or missing. Note that debiasing is not performed on the training set here.

[0122] Further, follow Figure 2 The prediction process of the causal graph shown in (b) evaluates the answer results in the real world and optimizes the parameters by minimizing the binary cross entropy loss between the true answer and the predicted probability. The calculation formula of the loss function of this task is as follows:

[0123] in, is the first loss function.

[0124] Then, for the learning tendency driven path prediction task and the problem representation influenced path prediction task, the binary cross entropy loss function is also used to measure the prediction accuracy. The calculation formula is as follows:

[0125] in, is the second loss function, is the third loss function.

[0126] The loss function of the answer result prediction model is obtained based on the first loss function, the second loss function, and the third loss function. The calculation formula is as follows:

[0127] in, is the first loss function, is the second loss function, is the third loss function, is the first hyperparameter, is the second hyperparameter.

[0128] In this example, a pre-defined answer prediction model is constructed and trained using datasets of different student information, different questions, and different answer results to generate an answer prediction model. Optimization using a binary cross-entropy loss function improves prediction accuracy, thereby supporting the development of personalized learning plans, improving student learning outcomes, and reducing the waste of teaching resources.

[0129] The knowledge tracking method based on counterfactual personalized enhancement provided in the embodiments of the present application can be executed by a knowledge tracking device based on counterfactual personalized enhancement. In the embodiments of the present application, the knowledge tracking method based on counterfactual personalized enhancement is executed by a knowledge tracking device based on counterfactual personalized enhancement as an example to illustrate the knowledge tracking device based on counterfactual personalized enhancement provided in the embodiments of the present application.

[0130] The present application also provides a knowledge tracking device based on counterfactual personalized enhancement, such as Figure 4 As shown, the knowledge tracking device based on counterfactual personalized enhancement includes: an acquisition module 410 , a first construction module 420 , a second construction module 430 , a first calculation module 440 , a second calculation module 450 and a prediction module 460 .

[0131] The acquisition module 410 is used to obtain student information, questions and answers, and define knowledge tracking, which is to predict the probability of students' correct answers to questions based on their answers to historical questions and student information. A first construction module 420 is configured to construct a knowledge tracing causal graph based on knowledge tracing, wherein the knowledge tracing causal graph includes a node set and a path set, wherein the node set includes students, questions, knowledge states, and answer results, and the path set includes a learning tendency driving path, a question representation influencing path, and a knowledge state mapping path; A second construction module 430 is configured to construct an answer result prediction model based on the knowledge tracing causal graph, wherein the answer result prediction model includes a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question representation impact path sub-model, and a fusion output sub-model; A first calculation module 440 is configured to calculate a path total effect, a path direct effect, and a path indirect effect based on the knowledge tracing causal graph; A second calculation module 450 is configured to obtain a debiasing formula for counterfactual prediction of the student's answer to the question based on the total path effect, the direct path effect, the indirect path effect, and the answer result prediction model; The prediction module 460 is used to obtain the counterfactual prediction value of the student's answer to the question based on the counterfactual prediction debiasing formula.

[0132] According to the knowledge tracking method based on counterfactual personalized enhancement provided in the embodiment of the present application, by obtaining student information, questions and answer results, and defining the knowledge tracking problem based on this, combined with the counterfactual prediction debiasing formula, by calculating the total path effect, path direct effect and path indirect effect of the knowledge tracking causal graph, deep learning and causal inference technology are comprehensively utilized to explore students' local learning tendencies and global problem representations, and a method that can support different knowledge tracking models is designed to capture students' local learning tendencies and global problem representations, and use counterfactual inference to effectively alleviate the global bias in the knowledge tracking task, thereby enhancing students' personalized performance.

[0133] The knowledge tracking device based on counterfactual personalized enhancement provided by the embodiment of the present application can achieve Figures 1 to 3 To avoid repetition, the various processes implemented in the embodiment of the knowledge tracking method based on counterfactual personalized enhancement are not described here.

[0134] In some embodiments, as Figure 5 As shown, an embodiment of the present application also provides an electronic device 500, including a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. When the program is executed by the processor 501, each process of the above-mentioned embodiment of the knowledge tracking method based on counterfactual personalized enhancement is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0135] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0136] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned knowledge tracking method embodiment based on counterfactual personalized enhancement are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0137] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0138] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned knowledge tracking method based on counterfactual personalized enhancement.

[0139] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0140] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned knowledge tracking method embodiment based on counterfactual personalized enhancement, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0141] It should be understood that the chip mentioned in the embodiments of the present application can also be called a device-level chip, a device chip, a chip device, or an on-chip device chip, etc.

[0142] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0143] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course, by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the knowledge tracking method based on counterfactual personalized enhancement of each embodiment of the present application.

[0144] In the description of this application, "first feature" and "second feature" may include one or more such features.

[0145] In the description of this application, “plurality” means two or more.

[0146] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

[0147] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0148] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and intent of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A knowledge tracking method based on counterfactual personalized enhancement, characterized in that: The method comprises: Obtain student information, questions, and answers, and define knowledge tracking, which is to predict the probability of a student's answer to a question being correct based on the student's answers to historical questions and the student information; Constructing a knowledge tracking causal graph based on knowledge tracking, wherein the knowledge tracking causal graph includes a node set and a path set, wherein the node set includes students, questions, knowledge states, and answer results, and the path set includes a learning tendency driving path, a question representation influencing path, and a knowledge state mapping path; Constructing an answer result prediction model based on the knowledge tracing causal graph, the answer result prediction model includes a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question representation impact path sub-model and a fusion output sub-model; Calculate the total path effect, the direct path effect, and the indirect path effect based on the knowledge tracing causal graph; Based on the total path effect, the direct path effect, the indirect path effect and the answer result prediction model, a counterfactual prediction debiasing formula for the student's answer result to the question is obtained; Based on the counterfactual prediction debiasing formula, the counterfactual prediction value of the student's answer to the question is obtained.

2. The knowledge tracking method based on counterfactual personalized enhancement according to claim 1 is characterized in that: The learning tendency driving path is student-answer result, which is used to describe the direct impact of student cognition on the answer result. The problem representation influence path is question-answer result, which is used to describe the direct impact of the question's own attributes on the answer result. The knowledge state mapping path is student and question-knowledge state-answer result.

3. The knowledge tracking method based on counterfactual personalized enhancement according to claim 2 is characterized in that: The knowledge state mapping path includes an ability-problem matching path and an answer result prediction path. The ability-problem matching path is student and problem-knowledge state, which is used to describe the decisive effect of the matching degree between student ability and problem on the knowledge state. The answer result prediction path is knowledge state-answer result, which is used to describe the predictive effect of knowledge state on answer result.

4. The knowledge tracking method based on counterfactual personalized enhancement according to claim 1 is characterized in that: The learning tendency driven path sub-model includes a first fully connected network and a first multi-layer perceptron, which are used to obtain the learning tendency driven path prediction results. The knowledge state mapping path sub-model is a knowledge tracking model. The problem representation influence path sub-model includes a second fully connected network and a second multi-layer perceptron, which are used to obtain the problem representation influence path prediction results. The fusion output sub-model includes a third multi-layer perceptron, which is used to obtain the predicted value of the student's answer to the question.

5. The knowledge tracking method based on counterfactual personalized enhancement according to claim 4 is characterized in that: The calculation formula of the fusion output sub-model is as follows: ； in, In fusion state, is the fusion function, is the intermediate output state of the learning tendency driving pathway sub-model, The knowledge mastery state of the knowledge state mapping path sub-model, The intermediate output state of the problem representation impacts the path sub-model, Indicates splicing, is a trainable matrix, is the trainable bias vector, is the predicted value of the answer to the question, It is the calculation value of the answer to the real-world question, and MLP is the third multi-layer perceptron.

6. The knowledge tracking method based on counterfactual personalized enhancement according to claim 1 is characterized in that: The calculation formula of the counterfactual prediction debiasing formula is as follows: ； in, is the predicted value of the answer to the question, Predicting outcomes for learning propensity-driven pathways, The problem characterization affects the path prediction results, TIE is the path indirect effect, NDE is the path direct effect, The weight of the predicted results of the learning tendency driving path, The weight that affects the path prediction result for the problem representation, is a hyperparameter, is the counterfactual prediction value of the student's answer to the question, S is the student, Q is the question, and Y is the student's answer result. is the local learning tendency, t is the time step, is the global question representation of the question answered at time step t+1.

7. The knowledge tracking method based on counterfactual personalized enhancement according to claim 1 is characterized in that: The training process of the answer result prediction model includes: Construct a prediction model for preset answer results; Obtain different student information, different questions, and different answer results as a data set; Based on the loss function, a preset answer result prediction model is trained according to the data set to obtain the answer result prediction model, wherein the loss function is constructed based on the binary cross entropy loss function.

8. A knowledge tracking device based on counterfactual personalized enhancement, implemented using the knowledge tracking method based on counterfactual personalized enhancement according to any one of claims 1 to 7, characterized in that: The device comprises: An acquisition module is used to obtain student information, questions, and answer results, and define knowledge tracking, which is to predict the probability of a student's answer to a question being correct based on the student's answer results to historical questions and student information; A first construction module is configured to construct a knowledge tracing causal graph based on knowledge tracing, wherein the knowledge tracing causal graph includes a node set and a path set, wherein the node set includes students, questions, knowledge states, and answer results, and the path set includes a learning tendency driving path, a question representation influencing path, and a knowledge state mapping path; A second construction module is configured to construct an answer result prediction model based on the knowledge tracing causal graph, wherein the answer result prediction model includes a learning tendency driving path sub-model, a knowledge state mapping path sub-model, a question representation impact path sub-model, and a fusion output sub-model; A first calculation module is used to calculate the total effect of the path, the direct effect of the path and the indirect effect of the path based on the knowledge tracing causal graph; A second calculation module is used to obtain a counterfactual prediction debiasing formula for the student's answer to the question based on the total path effect, the direct path effect, the indirect path effect and the answer result prediction model; The prediction module is used to obtain the counterfactual prediction value of the student's answer to the question based on the counterfactual prediction debiasing formula.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the knowledge tracking method based on counterfactual personalized enhancement is implemented as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the knowledge tracking method based on counterfactual personalized enhancement is implemented as described in any one of claims 1 to 7.

Images