Data analysis method and system for collaborating student learning based on intelligent agent

Abstract

The invention relates to a data analysis method and system for collaborating student learning based on an intelligent agent, and belongs to the technical field of data analysis. The method comprises the following steps: continuously acquiring multi-modal data generated in a learning interaction process between a student and an AI agent; calculating initial scores of the n evaluation indexes at a plurality of moments; introducing a time attenuation factor, and calculating target scores of the n evaluation indexes at the t moment in combination with the initial scores of the n evaluation indexes at the multiple moments; the target score represents a comprehensive score of the evaluation index after time decay at the t moment; and based on the target scores of the n evaluation indexes at the moment t, determining a learning analysis result of the student at the moment t, and based on the learning analysis results of continuous q moments, outputting a data analysis curve of the student. Through the above mode, comprehensive tracking of real-time performance and long-term trend of students can be realized, dynamic nature and timeliness are provided, and hysteresis and limitation of an existing static evaluation system are overcome.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a data analysis method and system based on agent-based collaborative student learning. Background Technology

[0002] With the rapid development of artificial intelligence (AI) technology, more and more fields are beginning to engage with and utilize it. For example, AI technology has now penetrated the education field, reshaping students' learning patterns and experiences. From completing daily assignments to retrieving information, from breaking down complex problems to personalized knowledge organization, AI tools have become collaborative partners in the learning process for an increasing number of students.

[0003] Research has revealed a lack of effective tracking and evaluation of students' AI usage within the current education system. This leads to student dependence on AI and an uncontrollable learning process. Teachers struggle to obtain real-time, accurate student learning data, and evaluation methods remain primarily outcome-based, failing to reflect students' depth of thought, questioning abilities, and innovation during AI interactions. In other words, the existing education system's mechanisms for tracking and evaluating student AI usage are severely lagging, particularly at the data evaluation level. These mechanisms remain limited to conventional static indicators, failing to comprehensively and dynamically reflect students' true learning status and the effectiveness of AI collaboration. Current evaluation indicators primarily focus on outcome-based data, such as the accuracy rate and completeness of answers when students use AI tools to complete assignments. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, this invention provides a data analysis method and system for student learning based on agent-based collaborative learning.

[0005] In a first aspect, embodiments of this application provide a data analysis method for student learning based on intelligent agent collaboration, comprising: continuously acquiring multimodal data generated during the learning interaction between a student and an AI agent; calculating initial scores of n evaluation indicators at multiple times based on the multimodal data; n being a positive integer greater than 0; introducing a time decay factor and combining it with the initial scores of the n evaluation indicators at multiple times to calculate the target scores of the n evaluation indicators at time t; time t being the last time among the multiple times; the target score representing the comprehensive score of the evaluation indicators at time t after time decay; determining the student's learning analysis result at time t based on the target scores of the n evaluation indicators at time t, and outputting the student's learning analysis curve based on the learning analysis results of q consecutive times; q being a positive integer greater than 0.

[0006] In an optional implementation of the first aspect, determining the student's learning analysis result at time t based on the target scores of the n evaluation indicators at time t includes: performing a weighted calculation based on the target scores of the n evaluation indicators at time t and the weights corresponding to each evaluation indicator to obtain the student's learning analysis result at time t.

[0007] In an optional implementation of the first aspect, the method further includes: when the target score of a first evaluation indicator is lower than a first threshold within a first set period, an adaptive update of the weight corresponding to the first evaluation indicator is triggered; wherein the first evaluation indicator is any one of the n evaluation indicators.

[0008] In an optional implementation of the first aspect, the method further includes: when the target score of the second evaluation indicator is lower than the second threshold within a second set period, the generation of a sub-indicator corresponding to the second evaluation indicator is triggered; based on the target score of the second evaluation indicator, the second threshold, and a preset adjustment coefficient, the weight of the sub-indicator corresponding to the second evaluation indicator is determined; wherein the second evaluation indicator is any one of the n evaluation indicators.

[0009] In one optional implementation of the first aspect, the number of AI agents is M; M is a positive integer greater than 1; the M AI agents have different functional types; the method further includes: obtaining the target scores of n evaluation indicators corresponding to the M AI agents at time t; determining the student's multi-agent learning analysis result at time t based on the weights of the M AI agents and the target scores of the n evaluation indicators corresponding to the M AI agents at time t; and outputting the student's multi-agent data analysis curve based on the multi-agent learning analysis results over q consecutive time periods.

[0010] In one optional implementation of the first aspect, the weights of the M AI agents are dynamically adjusted by the following steps: determining the systematic deviation value between any two AI agents based on the target scores of n evaluation indicators corresponding to the M AI agents at time t; and dynamically adjusting the weights of the M AI agents based on the systematic deviation value between any two AI agents.

[0011] In an optional implementation of the first aspect, the method further includes: extracting the student's behavioral data from the multimodal data; determining, based on the student's behavioral data and a set anomaly indicator, whether the student exhibits abnormal behavior during learning interaction with the AI ​​agent; wherein the set anomaly indicator is based on dynamic changes in historical behavior.

[0012] In an optional implementation of the first aspect, determining whether the student exhibits abnormal behavior during learning interaction with the AI ​​agent based on the student's behavioral data and a set anomaly indicator includes: if the student's behavioral data at time t is greater than the set anomaly indicator, determining that the student exhibits abnormal behavior at time t; introducing a sliding window mechanism to determine the probability of abnormal behavior occurring within the sliding window; and automatically generating a personalized intervention mechanism if the probability of abnormal behavior occurring within the sliding window is greater than a preset sensitivity threshold.

[0013] In one alternative implementation of the first aspect, the personalized intervention mechanism includes at least one of prompting the teacher's terminal device, initiating task intervention by the AI ​​agent itself, or imposing behavioral restrictions on the student.

[0014] Secondly, this application provides a data analysis system for student learning based on intelligent agent collaboration, comprising: an acquisition module for continuously acquiring multimodal data generated during the learning interaction between the student and the AI ​​agent; a first calculation module for calculating the initial scores of n evaluation indicators at multiple times based on the multimodal data, where n is a positive integer greater than 0; a second calculation module for introducing a time decay factor and, in conjunction with the initial scores of the n evaluation indicators at multiple times, calculating the target scores of the n evaluation indicators at time t, where time t is the last of the multiple times; the target score represents the comprehensive score of the evaluation indicators at time t after time decay; and a data evaluation module for determining the student's learning analysis result at time t based on the target scores of the n evaluation indicators at time t, and outputting the student's learning analysis curve based on the learning analysis results at q consecutive times, where q is a positive integer greater than 0.

[0015] The beneficial effects of this invention include: First, by collecting multimodal data during the interaction process, it breaks through the limitation of traditional evaluation relying solely on text results, and can simultaneously capture students' cognitive processes, operational behaviors, and feedback attitudes, making the evaluation dimensions more comprehensive and the results closer to students' actual learning status.

[0016] Secondly, by employing a time decay mechanism to dynamically score students' performance across various learning indicators, this approach achieves comprehensive tracking of students' immediate performance and long-term trends. It possesses dynamism and timeliness, overcoming the lag and limitations of existing static evaluation systems. Specifically, it integrates historical and current performance across various evaluation indicators, enabling a comprehensive evaluation of both instantaneous behavior and long-term trends. This not only quantifies the dynamic performance of students across different indicators but also allows learning analysis results to adaptively reflect students' learning progress at different time points. This helps educators identify learning patterns, predict development trends, and provides data support for personalized instruction. Attached Figure Description

[0017] Figure 1 A flowchart illustrating the steps of a data analysis method for student learning based on agent-based collaborative learning, provided in an embodiment of the present invention; Figure 2 A flowchart illustrating the steps of another data analysis method for student learning based on agent-based collaborative learning provided in this embodiment of the invention; Figure 3 A flowchart illustrating the steps of another data analysis method for student learning based on agent collaboration provided in this embodiment of the invention; Figure 4 This is a block diagram of a data analysis system for student learning based on agent collaboration, provided in an embodiment of the present invention. Figure 5 This is a module block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0019] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0020] Research has revealed a lack of effective tracking and evaluation of students' AI usage within the current education system. This leads to student dependence on AI and an uncontrollable learning process. Teachers struggle to obtain real-time, accurate student learning data, and evaluation methods remain primarily outcome-based, failing to reflect students' depth of thought, questioning abilities, and innovation during AI interactions. In other words, the existing education system's mechanisms for tracking and evaluating student AI usage are severely lagging, particularly at the data evaluation level. These mechanisms remain limited to conventional static indicators, failing to comprehensively and dynamically reflect students' true learning status and the effectiveness of AI collaboration. Current evaluation indicators primarily focus on outcome-based data, such as the accuracy rate and completeness of answers when students use AI tools to complete assignments.

[0021] In view of the above problems, this application proposes the following embodiments to solve the above technical problems.

[0022] Please see Figure 1This application provides a data analysis method for student learning based on agent collaboration, including steps 101 to 104.

[0023] Step 101: Continuously acquire multimodal data generated during the learning interaction between students and the AI ​​agent.

[0024] It should be noted that an AI agent is an artificial intelligence model that can understand complex instructions, perform logical reasoning, autonomously break down tasks, and call upon tools to complete tasks, possessing autonomy, interactivity, and goal orientation.

[0025] The multimodal data generated in the above steps includes, but is not limited to, student questions, interaction rounds, feedback verification, dialogue duration, student modification history of answers, frequency of repeated queries, and labeled content.

[0026] The aforementioned multimodal data can be further subdivided into text-based data, behavioral data, and feedback data.

[0027] Textual data can refer to student questions, AI-generated answers, and student modifications to those answers. Behavioral data can refer to interaction rounds and frequency of repeated queries. Feedback data refers to labeled content and feedback verification.

[0028] In the specific calculation process, multimodal data can be converted into corresponding feature vectors to facilitate subsequent data analysis.

[0029] It should be noted that the aforementioned multimodal data is continuously acquired during the learning interaction between students and the AI ​​agent.

[0030] Step 102: Based on multimodal data, calculate the initial scores of n evaluation indicators at multiple time points.

[0031] Where n is a positive integer greater than 0. Evaluation metrics may include, but are not limited to, the quality of student questions, the depth of interaction, creative expression, fluency of language expression, and the appropriateness of AI tool use, etc.

[0032] That is, in this step, based on the collected multimodal data, n evaluation indicators are quantified in real time to obtain the initial score of each evaluation indicator at multiple time points.

[0033] It should be noted that the scoring method here can be different depending on the evaluation indicators. For example, the fluency of language expression can be determined by identifying whether there are typos, incoherent sentences, or incorrect punctuation in the input content, and then outputting a score. The depth of interaction can be determined by the number of times the student and the agent ask each other questions on the same issue. This application does not limit the specific scoring method; the score range can be set to 0-100 points, or 0-10 points, etc., and this application does not impose any limitations.

[0034] The set of n evaluation indicators can be represented as: ;in, Represents the set of evaluation indicators. This indicates the first evaluation indicator. This indicates the second evaluation indicator. This represents the nth evaluation indicator.

[0035] Accordingly, the initial score can be expressed as: This parameter represents the initial score for the j-th evaluation metric at time t.

[0036] Step 103: Introduce a time decay factor and combine the initial scores of n evaluation indicators at multiple times to calculate the target scores of the n evaluation indicators at time t.

[0037] Here, time t is the last time among multiple times.

[0038] It should be noted that by introducing a time decay mechanism, the initial score at time t is dynamically adjusted, so that the weight of recent learning behavior data is higher, and the weight of long-term data decays over time, thereby generating the target score.

[0039] Specifically, to avoid analysis based solely on instantaneous performance, this embodiment introduces a time decay memory mechanism, using a decay factor... Weighting historical performance increases the impact of recent behavior while gradually decreasing the impact of long-term performance. The specific calculation formula is as follows: ; In the above formula, Indicates a historical time index. This represents the comprehensive score of the j-th evaluation index after time decay at time t (corresponding to the target score mentioned above). Indicates time The initial score of the j-th evaluation indicator.

[0040] Step 104: Based on the target scores of n evaluation indicators at time t, determine the student's learning analysis results at time t, and based on the learning analysis results over q consecutive time points, output the student's data analysis curve.

[0041] Where q is a positive integer greater than 0.

[0042] Finally, based on the target score of each evaluation indicator at time t, the student's learning analysis results at time t are obtained, and the student's data analysis curve can be automatically plotted based on the learning analysis results of q consecutive time points.

[0043] The learning analysis results of the students at time t can refer to the overall performance across all n evaluation indicators. Alternatively, the learning analysis results at time t can be the learning analysis results for each individual evaluation indicator; that is, the learning analysis results for each evaluation indicator can be determined based on the score of each target indicator.

[0044] In summary, the data analysis method for student learning based on agent-based collaborative learning provided in this application has the following beneficial effects: First, by collecting multimodal data during the interaction process, it breaks through the limitation of traditional evaluation relying solely on text results. It can simultaneously capture students' cognitive processes, operational behaviors, and feedback attitudes, making the evaluation dimensions more comprehensive and the results closer to students' actual learning status.

[0045] Secondly, by employing a time decay mechanism to dynamically score students' performance across various learning indicators, this approach achieves comprehensive tracking of students' immediate performance and long-term trends. It possesses dynamism and timeliness, overcoming the lag and limitations of existing static evaluation systems. Specifically, it integrates historical and current performance across various evaluation indicators, enabling a comprehensive evaluation of both instantaneous behavior and long-term trends. This not only quantifies the dynamic performance of students across different indicators but also allows learning analysis results to adaptively reflect students' learning progress at different time points. This helps educators identify learning patterns, predict development trends, and provides data support for personalized instruction.

[0046] Optionally, step 104 above, which determines the student's learning analysis result at time t based on the target scores of n evaluation indicators at time t, may specifically include: performing a weighted calculation based on the target scores of the n evaluation indicators at time t and the weights corresponding to each evaluation indicator to obtain the student's learning analysis result at time t.

[0047] Initially, each evaluation indicator can be assigned a different initial weight based on its importance. All weights are summed to 1. The student's learning analysis results at time t can then be derived using the following formula: ; In the above formula, This represents the student's learning analysis results at time t (the overall performance under various evaluation indicators). This represents the weight of the j-th evaluation indicator. This represents the set of evaluation indicators.

[0048] In summary, assigning differentiated weights to different indicators can reflect the guiding nature of educational goals. For example, higher weights can be given to evaluation indicators such as "creative expression" and "fluency of language expression," making the learning analysis results more aligned with teaching needs. Furthermore, the weighted comprehensive learning analysis results avoid the one-sidedness of single-indicator evaluation and solve the decision-making difficulties caused by excessive data volume when multiple indicators are used in parallel, providing educators with a comprehensive yet concise basis for judgment.

[0049] Optionally, based on the above embodiments, the method further includes: when the target score of a first evaluation indicator is lower than a first threshold within a first set period, the weight corresponding to the first evaluation indicator is adaptively updated; wherein, the first evaluation indicator is any one of n evaluation indicators.

[0050] The first threshold and the first set period can both be set according to requirements.

[0051] Among them, if the target score of the first evaluation indicator Less than If this happens, an adaptive weight update is triggered. The above uses the j-th evaluation metric as the first evaluation metric. This represents the first threshold.

[0052] This application provides an adaptive weight update method, which can be referred to by the following formula: ; in, This represents the current weight of the j-th evaluation indicator. This represents the updated weight of the j-th evaluation indicator. Indicates the learning rate. This represents the current weight of the k-th evaluation indicator. and Both indicate adjustment signals, when Less than , A value greater than 0 increases the weight of the indicator; otherwise, a value of zero or negative is used to reduce the weight; k indicates that when summing the denominators, all evaluation indicator numbers are traversed.

[0053] For example, if the score of the independent thinking ability indicator is below 60 points (the first threshold) for a first set period, the weight of this indicator in the comprehensive evaluation will be automatically increased to strengthen the monitoring and guidance of students' ability in this area.

[0054] It is evident that by setting up an automatic triggering mechanism, the evaluation system can self-adjust based on students' actual performance, avoiding the static and rigid problems of weight settings. In contrast, traditional evaluation systems rely on educators' experience and manual operation for weight adjustments, resulting in delays and subjectivity. This mechanism achieves automatic weight adaptation through preset rules, improving the timeliness of adjustments and reducing manual maintenance costs, making the evaluation system more practical and operable.

[0055] Optionally, based on the above embodiments, the method further includes: when the target score of the second evaluation indicator is lower than the second threshold within a second set period, the generation of the sub-indicator corresponding to the second evaluation indicator is triggered; the weight of the sub-indicator corresponding to the second evaluation indicator is determined based on the target score of the second evaluation indicator, the second threshold, and the preset adjustment coefficient; wherein, the second evaluation indicator is any one of the n evaluation indicators.

[0056] The second threshold and the second set period can both be set according to requirements. The second threshold can be less than the first threshold in the aforementioned embodiment.

[0057] The above embodiments can automatically generate new sub-indicators after weaknesses are identified.

[0058] For example, when the "question quality" metric remains below the second threshold At that time, the system automatically introduces the "question guidance" sub-indicator.

[0059] For example, when the "reasonable use of AI tools" indicator fails to meet the standard, sub-indicators such as "inappropriate timing of use" and "incorrect tool selection" can be automatically introduced.

[0060] The weighting of sub-indicators can be set with reference to the following formula: ; in, Indicates the weight of the generated sub-indicators; This is a preset adjustment coefficient used to control the influence of newly added sub-indicators.

[0061] It should be noted that the above method ensures that the evaluation system can not only dynamically adjust the weights of existing indicators but also expand into new analytical dimensions, thereby more comprehensively depicting students' learning status. Specifically, when a core indicator consistently performs poorly, generating sub-indicators can break down the general evaluation dimension into more specific ability elements, helping to pinpoint the root cause of the problem. The generation of sub-indicators is essentially a dynamic refinement of the evaluation system, transforming vague notions of "weak ability" into specific "defects in certain aspects." This helps educators determine students' specific shortcomings in a particular ability dimension based on sub-indicator data. Therefore, the above method enhances the adaptability and scalability of the evaluation system, as well as strengthens the granular guidance for the learning process.

[0062] Optionally, the number of AI agents is M; M is a positive integer greater than 1; the M AI agents have different functional types.

[0063] For example, the M AI agents can be categorized by function type, including exam answering agents, writing assistance agents, sentiment analysis agents, and homework correction agents.

[0064] For M AI agents, embodiments of this application also provide a data evaluation method based on multiple agents. Please refer to [link to relevant documentation]. Figure 2 The method specifically includes steps 201 to 202.

[0065] Step 201: Obtain the target scores of n evaluation metrics for M AI agents at time t.

[0066] Step 202: Based on the weights of each of the M AI agents and the target scores of the n evaluation indicators corresponding to the M AI agents at time t, determine the student's multi-agent learning analysis results at time t, and output the student's multi-agent data analysis curve based on the learning analysis results over q consecutive time steps.

[0067] For the target scores of the n evaluation indicators for each AI agent at time t, please refer to the explanations in steps 101 to 104 above, which will not be repeated here.

[0068] When obtaining the target scores of n evaluation metrics for M AI agents at time t, standardization can be performed first to ensure that the scores of different AI agents are comparable.

[0069] The results of the multi-agent learning analysis of the students at time t can be referenced by the following formulas, including: ; in, This represents the results of the student's multi-agent learning analysis at time t; This represents the weight of the m-th AI agent; This represents the target score of the j-th evaluation metric corresponding to the m-th AI agent; This represents the weight of the j-th evaluation indicator.

[0070] In reality, students often use multiple AI tools with different functions to meet their learning needs (such as simultaneously using translation software, grammar checking, and literature analysis tools). This solution incorporates multiple agents into the data analysis system, making the application scenarios more closely aligned with actual learning behaviors. It overcomes the limitations of evaluating a single AI agent and achieves a comprehensive assessment of the ability to use AI tools.

[0071] Optionally, the weights of the M AI agents can be dynamically adjusted through the following steps: determining the systematic deviation value between any two AI agents based on the target scores of the n evaluation indicators corresponding to the M AI agents at time t; and dynamically adjusting the weights of the M AI agents based on the systematic deviation value between any two AI agents.

[0072] The formula for calculating the systematic deviation value can be found as follows: ; in, Indicates the systematic deviation value. This represents the target score of the j-th evaluation metric corresponding to the m-th AI agent; This represents the target score of the j-th evaluation metric corresponding to the n-th AI agent. The larger the value of , the greater the systematic deviation between the m-th AI agent and the n-th AI agent. Then, for AI agents with significant systematic deviations, their weight allocation is dynamically optimized according to the direction and degree of deviation (e.g., reducing the weight of AI agents with significant deviations from the majority of agents).

[0073] The above methods can identify and address systematic biases among AI agents, reducing evaluation distortion caused by differences in the characteristics of different AI agents. When an individual AI agent's evaluation is abnormal due to algorithmic limitations or data bias, systematic bias analysis can automatically identify and reduce its weight, minimizing the interference of abnormal data on the overall evaluation and improving the stability of the evaluation system.

[0074] Optionally, embodiments of this application also provide an anomaly detection mechanism; please refer to [link to relevant documentation]. Figure 3 The method may further include steps 301 to 302.

[0075] Step 301: Extract student behavioral data from multimodal data.

[0076] Behavioral data may include, but is not limited to, the number of questions asked, the method of answer submission, the duration of operation, etc.

[0077] Step 302: Based on the student's behavioral data and the set abnormal indicators, determine whether there is any abnormal behavior in the student's learning interaction with the AI ​​agent.

[0078] The types of abnormal indicators that can be set can be, but are not limited to, plagiarism coefficient, insufficient follow-up question coefficient, and task completion time.

[0079] Among them, the plagiarism coefficient is used to measure the similarity between the student's submitted answer and the reference answer output by the AI ​​agent; the insufficient follow-up question coefficient is used to measure the proportion of students who did not ask supplementary questions; and the time-out dependence coefficient is used to measure the time spent continuously relying on AI to complete the task.

[0080] The thresholds for the abnormal indicators are set based on dynamic changes in historical behavior.

[0081] That is, a dynamic threshold mechanism is introduced to adaptively adjust the anomaly detection based on historical performance. See the following formula for details: ; in, Representing behavioral data An adaptive threshold at time t; and Representing behavioral data The moving mean and standard deviation; This represents the adjustment coefficient, which controls the threshold sensitivity.

[0082] The following are some methods for identifying abnormal behavior: ; This indicates the result of abnormal behavior judgment; the adaptive threshold represents behavioral data when it equals 1. An anomaly has occurred. Conversely, if the anomaly is not present, it is not an anomaly.

[0083] As can be seen, by focusing on behavioral data rather than outcome data, the above embodiments can identify potential learning problems (such as over-reliance on AI, perfunctory learning, etc.) earlier, avoiding the lag of judging solely based on the final result. Furthermore, this approach adapts to individual behavioral differences and developmental changes: it dynamically adjusts abnormal indicators based on historical behavior, fully considering the differences in learning habits among different students (e.g., some students take longer to think, while others ask questions more frequently), and it can also adapt to the natural evolution of student behavior patterns, solving the problem of applying a fixed threshold to all students in a "one-size-fits-all" manner. Real-time identification of abnormal behavior allows educators to intervene in the early stages of problems. For example, when the system detects that a student suddenly copies AI answers frequently, it can promptly remind them to focus on independent thinking, preventing bad learning habits from becoming entrenched.

[0084] Optionally, step 302 above determines whether there is abnormal behavior in the student's learning interaction with the AI ​​agent based on the student's behavioral data and the set abnormality indicators, including: if the student's behavioral data at time t is greater than the set abnormality indicators, it is determined that the student has abnormal behavior at time t; a sliding window mechanism is introduced to determine the probability of abnormal behavior occurring within the sliding window; if the probability of abnormal behavior occurring within the sliding window is greater than a preset sensitivity threshold, a personalized intervention mechanism is automatically generated.

[0085] The aforementioned sliding window can be specifically set to 30 minutes, the last five interactions, etc., and this application does not impose any limitations. The aforementioned preset sensitivity threshold can also be set according to actual circumstances.

[0086] Introducing a sliding window mechanism to determine the sliding window The probability of abnormal behavior in the internal indicator k You can refer to the following formula: ;in, Indicates a time index. Indicates time The results of the abnormal behavior judgment.

[0087] Optionally, personalized intervention mechanisms include at least one of prompting the teacher's terminal device, initiating task intervention by the AI ​​agent itself, or imposing behavioral restrictions on the student.

[0088] The system includes prompting teachers' devices to display reports and trends of abnormal student behavior. Task interventions that activate the AI ​​agent itself can include requesting additional questions, requiring teachers to answer again, or delaying AI use. Restrictions on student behavior can include limiting AI usage time or specific functions.

[0089] In summary, calculating anomaly probabilities through a sliding window, rather than directly intervening based on a single abnormal behavior, effectively filters out occasional behavioral fluctuations (such as anomalies caused by a single misoperation), reduces unnecessary interventions, and makes the system more robust. Intervention is only deemed necessary when abnormal behaviors occur in clusters within a certain period (probability exceeding a threshold), ensuring that intervention measures target persistent, trend-based abnormal patterns rather than random events. The sliding window mechanism can both promptly capture recently concentrated abnormal behaviors (ensuring timely intervention) and avoid misjudgments due to momentary fluctuations in individual behaviors (ensuring the rationality of intervention), achieving a balance between sensitivity and stability.

[0090] Furthermore, when the number of AI agents is M, the feedback from each agent can be used as a weighted reference for anomaly detection. Additionally, different set anomaly indicators can be used to generate a total anomaly score; however, this application does not limit the scope of these scores.

[0091] Please see Figure 4 Based on the same inventive concept, embodiments of this application also provide a data analysis system 400 based on agent-based collaborative student learning, comprising: The acquisition module 401 is used to continuously acquire multimodal data generated during the learning interaction between students and AI agents.

[0092] The first calculation module 402 is used to calculate the initial scores of n evaluation indicators at multiple times based on the multimodal data; n is a positive integer greater than 0.

[0093] The second calculation module 403 is used to introduce a time decay factor and, in combination with the initial scores of the n evaluation indicators at multiple times, calculate the target score of the n evaluation indicators at time t; where time t is the last time among the multiple times; the target score represents the comprehensive score of the evaluation indicators after time decay at time t.

[0094] The data evaluation module 404 is used to determine the student's learning analysis result at time t based on the target scores of the n evaluation indicators at time t, and to output the student's learning analysis curve based on the learning analysis results at q consecutive times; where q is a positive integer greater than 0.

[0095] Please see Figure 5 Based on the same inventive concept, this application provides a module frame for an electronic device 500 that applies the above-described method. The electronic device 500 includes: at least one processor 501 (… Figure 5 (Only one is shown in the diagram), memory 502, computer program 503 stored in memory 502 and executable on at least one processor 501, processor 501 executing computer program 503 to implement the steps of the methods in any of the foregoing embodiments.

[0096] The electronic device 500 can be a mobile phone, tablet computer, server, personal computer, laptop computer, etc.

[0097] Those skilled in the art will understand that Figure 5 This is merely an example of electronic device 500 and does not constitute a limitation on electronic device 500. It may include more or fewer components than shown, or combine certain components, or use different components.

[0098] The processor 501 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0099] In some embodiments, the memory 502 may be an internal storage unit of the electronic device 500, such as a hard disk or memory of the electronic device 500. In other embodiments, the memory 502 may be an external storage device of the electronic device 500, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 500. Furthermore, the memory 502 may include both internal storage units and external storage devices of the electronic device 500.

[0100] It should be noted that the above-mentioned systems, devices, etc. are based on the same concept as the method embodiments of this application. The modules designed in the system, as well as the steps performed by the device and the resulting technical effects, can all be found in the method embodiments section, and will not be repeated here.

[0101] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0102] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0103] This application provides a computer program product that, when run on a mobile terminal, enables the mobile terminal to implement the steps described in the above-described method embodiments.

[0104] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a photographic device / electronic device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0105] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0106] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0108] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A data analysis method for student learning based on agent-based collaborative learning, characterized in that, include: Continuously acquire multimodal data generated during the learning interaction between students and AI agents; Based on the multimodal data, calculate the initial scores of n evaluation indicators at multiple time points; n is a positive integer greater than 0; A time decay factor is introduced, and the target score of the n evaluation indicators at time t is calculated by combining the initial scores of the n evaluation indicators at multiple times; time t is the last time among the multiple times; the target score represents the comprehensive score of the evaluation indicators after time decay at time t. Based on the target scores of the n evaluation indicators at time t, the student's learning analysis result at time t is determined, and based on the learning analysis results at q consecutive times, the student's learning analysis curve is output; q is a positive integer greater than 0.

2. The data analysis method for student learning based on agent-based collaborative learning according to claim 1, characterized in that, The determination of the student's learning analysis results at time t based on the target scores of the n evaluation indicators at time t includes: The student's learning analysis results at time t are obtained by weighting the target scores of the n evaluation indicators at time t and the weights corresponding to each evaluation indicator.

3. The data analysis method for student learning based on agent-based collaborative learning according to claim 2, characterized in that, The method further includes: If the target score of the first evaluation indicator is lower than the first threshold within the first set period, the weight corresponding to the first evaluation indicator will be adaptively updated. Wherein, the first evaluation index is any one of the n evaluation indices.

4. The data analysis method for student learning based on agent-based collaborative learning according to claim 2, characterized in that, The method further includes: If the target score of the second evaluation indicator is lower than the second threshold within the second set period, the generation of the sub-indicator corresponding to the second evaluation indicator will be triggered. Based on the target score of the second evaluation indicator, the second threshold, and the preset adjustment coefficient, the weight of the sub-indicator corresponding to the second evaluation indicator is determined. The second evaluation index is any one of the n evaluation indexes.

5. The data analysis method for student learning based on agent-based collaborative learning according to claim 1, characterized in that, The number of AI agents is M; M is a positive integer greater than 1; the M AI agents have different functional types; The method further includes: Obtain the target scores of n evaluation indicators corresponding to the M AI agents at time t; Based on the weights of the M AI agents and the target scores of the corresponding n evaluation indicators at time t, the multi-agent learning analysis results of the student at time t are determined, and the multi-agent data analysis curve of the student is output based on the multi-agent learning analysis results over q consecutive time periods.

6. The data analysis method for student learning based on agent-based collaborative learning according to claim 5, characterized in that, The weights of the M AI agents are dynamically adjusted through the following steps: Based on the target scores of n evaluation indicators corresponding to the M AI agents at time t, determine the systematic deviation value between any two AI agents; The weights of the M AI agents are dynamically adjusted based on the systematic deviation between any two AI agents.

7. The data analysis method for student learning based on agent-based collaborative learning according to claim 1, characterized in that, The method further includes: Extract the student's behavioral data from the multimodal data; Based on the student's behavioral data and the set abnormal indicators, determine whether the student exhibits any abnormal behavior during the learning interaction with the AI ​​agent; The abnormal indicators set are based on dynamic changes in historical behavior.

8. The data analysis method for student learning based on agent-based collaborative learning according to claim 7, characterized in that, The process of determining whether a student exhibits abnormal behavior during learning interaction with the AI ​​agent, based on the student's behavioral data and predefined anomaly indicators, includes: If the student's behavioral data at time t is greater than the set abnormality index, it is determined that the student has abnormal behavior at time t. A sliding window mechanism is introduced to determine the probability of abnormal behavior occurring within the sliding window; If the probability of abnormal behavior occurring within the sliding window exceeds a preset sensitivity threshold, a personalized intervention mechanism will be automatically generated.

9. The data analysis method for student learning based on agent-based collaborative learning according to claim 8, characterized in that, The personalized intervention mechanism includes at least one of the following: prompting the teacher's terminal device, initiating task intervention by the AI ​​agent itself, or restricting the student's behavior.

10. A data analysis system for student learning based on agent-based collaborative learning, characterized in that, include: The acquisition module is used to continuously acquire multimodal data generated during the learning interaction between students and the AI ​​agent; The first calculation module is used to calculate the initial scores of n evaluation indicators at multiple times based on the multimodal data; n is a positive integer greater than 0; The second calculation module is used to introduce a time decay factor and, in combination with the initial scores of the n evaluation indicators at multiple times, calculate the target score of the n evaluation indicators at time t; where time t is the last time among the multiple times; the target score represents the comprehensive score of the evaluation indicators after time decay at time t. The data evaluation module is used to determine the student's learning analysis result at time t based on the target scores of the n evaluation indicators at time t, and to output the student's learning analysis curve based on the learning analysis results at q consecutive times; where q is a positive integer greater than 0.

Images