Data analysis method for personalized diagnosis and intervention of student learning problems

Abstract

The invention provides a data analysis method for personalized diagnosis and intervention of student learning questions, and the method comprises the following steps: S1, collecting an answer event sequence of a target student in a target question, the answer event sequence comprising an operation type and a corresponding timestamp; and S2, performing process segmentation on the answer event sequence based on a time interval of adjacent answer events, a submission event, a jump event and a rollback event. According to the method, process segmentation, step alignment and abnormal evidence extraction are carried out on the answering event sequence in the answering process of the student, and the learning abnormity of the student is specifically positioned to the corresponding standard question solving step and the trigger evidence thereof, so that a structured learning question positioning result is formed; therefore, a learning analysis result is converted from a single result index into traceable and explainable process diagnosis information, coarse-grained judgment only depending on an answer result or an abnormal score is avoided, and usability and pertinence of learning problem analysis are improved.

Description

Technical Field

[0001] This invention relates to the field of learning data analysis technology, and in particular to a data analysis method for personalized diagnosis and intervention of students' learning problems. Background Technology

[0002] With the widespread adoption of online learning platforms, smart classrooms, and homework systems, the teaching process generates a large amount of data related to student learning behavior, including but not limited to homework and quiz answer records, answer time and number of retakes, learning resource clicks and viewing history, classroom interaction texts, and structured log data related to the learning process. Analyzing learning status based on this data has become an important technological direction for instructional management and learning support. In practical teaching applications, teachers or learning support systems usually need to carry out targeted interventions based on clear learning problem identification results. For example, they may explain specific knowledge points, provide error correction training for specific error types, or push relevant learning resources. If the analysis results lack interpretable diagnostic information and only provide abnormal scores, it is difficult to directly support subsequent personalized intervention decisions. It still needs to rely on manual analysis, which affects the efficiency and accuracy of teaching support. Therefore, a data analysis method for personalized diagnosis and intervention of students' learning problems is proposed. Summary of the Invention

[0003] In view of this, the present invention provides a data analysis method for personalized diagnosis and intervention of student learning problems, in order to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial option.

[0004] The technical solution of this invention is implemented as follows: a data analysis method for personalized diagnosis and intervention of student learning problems, comprising the following steps: S1. Collect the sequence of answer events of the target student in the target question, wherein the sequence of answer events includes the operation type and the corresponding timestamp; S2. Based on the time interval between adjacent response events and the submission event, jump event, and rollback event, the response event sequence is segmented into multiple response step segments arranged in chronological order, and the step segment features are statistically analyzed for each step segment. The step segment features include at least the duration, number of operations, and number of rollbacks. S3. Obtain the standard problem-solving process corresponding to the target problem. The standard problem-solving process consists of multiple standard steps arranged in sequence and their corresponding key operation sets. S4. For each answer step segment, calculate its alignment matching degree with each standard step based on the features of the step segment and the key operation set, and determine the step alignment sequence under the condition of satisfying the standard step order constraint. At the same time, identify the unaligned standard steps, the repeatedly aligned answer step segments, and the alignment relationships with inconsistent order. S5. Based on the step alignment sequence of step S4 and the duration and number of operations of each response step segment, extract abnormal evidence units that represent missing steps, sequence shifts, repeated modifications and abnormal time consumption. S6. Match the abnormal evidence unit with the preset learning problem type determination rule, and output a structured learning problem localization result containing the problem type, corresponding standard steps and triggering abnormal evidence; S7. Based on the structured learning problem localization results, retrieve and output intervention element data from the intervention mapping table, wherein the intervention mapping table uses problem type and corresponding standard step identifier as index items.

[0005] More preferably, each answer event further includes an event target identifier, which includes at least one of the question stem area, option area, draft area, and formula area; and the operation type includes at least two or more of the following: input, deletion, option switching, formula editing, submission, withdrawal, rollback, and question jump.

[0006] More preferably, the process segmentation in step S2 uses at least one of the following segmentation boundaries as the segment end boundary: (1) The time interval between adjacent response events is greater than the first time threshold; (2) A commit event occurs; (3) A question jump event occurs; (4) The number of backtracking events reaches the first threshold within the preset sliding time window; Furthermore, the first response event after the end boundary of the segment is used as the starting event for the next response step segment.

[0007] More preferably, the step fragment features further include at least one or more of the following features: (1) Count vectors of different event types within a segment; (2) The number of deletions and the number of undoes within a segment shall be considered as the number of modifications; (3) Intra-fragment backspan, whereby the backspan is the number of times the editing position is bounced back or the number of steps the interface view is rolled back due to the backspan event; (4) Net time between the first input event and the first submission event within the segment.

[0008] More preferably, each standard step of the standard problem-solving process is further configured with a step attribute identifier, and the step attribute identifier includes at least one of key step identifier and knowledge point identifier; The set of key operations includes at least one of the following: key intermediate quantity formation, key formula calling, key condition judgment, and key option elimination.

[0009] More preferably, step S4, determining the alignment sequence, includes a two-level determination process: The first determination process is as follows: For each response step segment, the matching degree is calculated based on the matching of the event type count vector of the response step segment with the key operation set to obtain a candidate alignment set; The second determination process: In the candidate alignment set, based on the order constraints of the standard step sequence, select the alignment combination that makes the overall alignment sequence consistent and maximizes the total matching degree; When a standard step is not aligned with any response step fragment, the standard step is marked as an unaligned standard step.

[0010] More preferably, step S5, which generates a set of abnormal evidence units, includes at least one or more of the following evidence units: E1. Evidence of missing steps: There are unaligned standard steps and the unaligned standard steps have critical step identifiers; E2, Sequence offset evidence: The number of inversion pairs of standard step identifiers in the step alignment sequence exceeds the second threshold; E3. Repeatedly modifying evidence: In the answer steps that are aligned with the same standard steps, delete the number of times the input is alternating with the deletion exceeds the third threshold, or the number of times the withdrawal is withdrawn exceeds the fourth threshold; E4. Evidence of abnormal time consumption: The duration of a response step segment aligned to the same standard step exceeds the fifth threshold of the reference duration of that standard step; E5. Evidence of backtracking: The cumulative backtracking span within a response step segment aligned to the same standard step exceeds the sixth threshold.

[0011] More preferably, the reference duration is calculated from a reference sample set, which is a set of durations of step segments of multiple reference students on the corresponding standard steps under the same target question; the reference duration is taken as the median or quantile value of the duration set.

[0012] More preferably, the structured learning problem localization result further includes evidence chain data, which includes at least the abnormal evidence unit identifier corresponding to the triggered problem type, the corresponding standard step identifier, and the start and end time range of the answer step segment corresponding to the abnormal evidence unit.

[0013] More preferably, the intervention mapping table maps problem types and corresponding standard step identifiers to an intervention element set, wherein the intervention element set includes at least two of the following: intervention resource identifier, intervention task type, and intervention intensity level; The intervention task type includes at least one or more of the following: explanation task, error correction task, consolidation exercise task, and retest task. The intervention intensity level is determined by the number of abnormal evidence units triggered or the severity level of the evidence.

[0014] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions: I. This invention segments, aligns, and extracts abnormal evidence from the sequence of events in a student's answer process, pinpointing the student's learning anomalies to the corresponding standard problem-solving steps and their triggering evidence. This results in a structured learning problem localization, transforming the learning analysis results from a single outcome indicator into traceable and interpretable process diagnostic information. This avoids relying solely on answer results or abnormal scores for coarse-grained judgments, thus improving the usability and relevance of learning problem analysis.

[0015] Second, this invention introduces problem type, corresponding standard steps, and evidence chain data into the learning problem location results, and establishes an intervention mapping relationship based on the problem type and standard steps, so that subsequent teaching interventions can be directly linked to specific knowledge points or problem-solving steps, reducing the reliance on manual question-by-question analysis and answering processes, which is conducive to improving the efficiency and consistency of personalized teaching intervention generation and enhancing the feasibility of the learning support process in actual teaching scenarios.

[0016] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the steps of the present invention. Detailed Implementation

[0019] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0020] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0021] like Figure 1 As shown, this embodiment of the invention provides a data analysis method for personalized diagnosis and intervention of student learning problems, including the following steps: Perform step S1 to collect the sequence of answer events of the target student in the target question; The answer event sequence is a set of events recorded chronologically during the student's answering of the target questions. Each answer event includes at least an operation type and a corresponding timestamp. Additionally, each answer event includes an event target identifier, indicating the interface area where the operation occurred. The event target identifier includes at least one of the following: question stem area, option area, draft area, and formula area. The operation type includes at least two or more of the following: input, deletion, option switching, formula editing, submission, withdrawal, rollback, and question jump, thus comprehensively reflecting the student's editing, modification, and rollback behaviors during the answering process.

[0022] After completing the collection of the response event sequence, step S2 is executed to perform process segmentation on the response event sequence; Based on the time interval between adjacent answer events, as well as submission events, question jump events, and rollback events, the answer event sequence is analyzed, and at least one of the following segmentation boundaries is used as the end boundary of the answer step segment: the time interval between adjacent answer events is greater than a first time threshold, a submission event is detected, a question jump event is detected, or the number of rollback events within a preset sliding time window reaches a first threshold. When any segmentation boundary is met, the current answer step segment is determined to end, and the first answer event after the segmentation boundary is used as the starting event of the next answer step segment, thereby dividing the complete answer event sequence into multiple answer step segments arranged in chronological order. For each answer step segment, its duration, number of operations, and number of rollbacks are recorded, and at least one of the following is further statistically analyzed: the count vector of different event types within the segment, the number of deletions, and the number of undoes as the number of modifications; the number of edit position jumps or interface view rollback steps caused by rollback events as the rollback span; and the net time taken between the first input event and the first submission event within the segment.

[0023] After generating the answer step fragment, execute step S3 to obtain the standard problem-solving process corresponding to the target question; The standard problem-solving process consists of multiple standard steps arranged in sequence. Each standard step is configured with a corresponding set of key operations and further configured with step attribute identifiers. The step attribute identifiers include at least one of key step identifiers and knowledge point identifiers. The set of key operations includes at least one of key intermediate quantity formation, key formula calling, key condition judgment, and key option elimination, so that the standard problem-solving process can establish a correspondence with the student's answer behavior.

[0024] Then, step S4 is executed, aligning the response step fragments obtained in step S2 with the standard steps obtained in step S3 one by one to generate a step alignment sequence. This alignment process includes a two-level determination process: For each response step segment, the matching degree is calculated based on the matching between the event type count vector of the response step segment and the key operation set corresponding to each standard step, and a candidate alignment set is obtained. Then, in the candidate alignment set, based on the order constraint of the standard steps in the standard problem-solving process, the alignment combination that makes the overall order of the step alignment sequence consistent and the sum of the matching degree is the largest is selected as the final step alignment sequence; When a standard step is not aligned with any answer step fragment, the standard step is marked as an unaligned standard step; when multiple answer step fragments are aligned to the same standard step, they are marked as a duplicate alignment relationship; when the order of the aligned standard steps is inconsistent with the order of the standard problem-solving process, they are marked as an inconsistent alignment relationship.

[0025] After obtaining the step alignment sequence, step S5 is executed. Based on the step alignment sequence and the duration and number of operations of each response step segment, abnormal evidence units are extracted to form an abnormal evidence unit set. Anomaly evidence units include at least one or more of the following: Evidence of missing steps arises when there are unaligned standard steps that have critical step identifiers; evidence of sequence offset arises when the number of inversion pairs of standard step identifiers in the step alignment sequence exceeds the second threshold; evidence of repeated modifications arises when the number of times the input is deleted from the response step segment aligned to the same standard step exceeds the third threshold or the number of times the withdrawal exceeds the fourth threshold; evidence of time consumption anomalies arises when the duration of the response step segment aligned to the same standard step exceeds the fifth threshold multiple of the reference duration of the standard step; and evidence of backtracking hesitation arises when the cumulative value of the backtracking span within the response step segment aligned to the same standard step exceeds the sixth threshold. The reference duration is calculated from the reference sample set, which is a set of the duration of step segments of multiple reference students on the corresponding standard steps under the same target question. The reference duration is taken as the median or quantile value of the set.

[0026] After extracting the abnormal evidence units, step S6 is executed to match the abnormal evidence units with the preset learning question type determination rules, and output the structured learning question localization result. The structured learning question localization result includes at least the question type, the corresponding standard steps, and the triggering abnormal evidence, and further includes evidence chain data. The evidence chain data includes at least the abnormal evidence unit identifier, the corresponding standard step identifier, and the start and end time range of the answer step segment corresponding to the abnormal evidence unit, so as to achieve traceable localization of the learning question.

[0027] Finally, step S7 is executed, based on the structured learning problem localization results obtained in step S6, to retrieve and output intervention element data from the intervention mapping table; The intervention mapping table uses problem type and corresponding standard step identifier as index items. The intervention element data includes at least two of the following: intervention resource identifier, intervention task type, and intervention intensity level. The intervention task type includes at least one or more of the following: explanation task, error correction task, consolidation exercise task, and retest task. The intervention intensity level is determined by the number of triggered abnormal evidence units or the severity level of evidence, thereby achieving targeted intervention output for students' specific learning problems.

[0028] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in the present invention, and these should all be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A data analysis method for personalized diagnosis and intervention of student learning problems, characterized in that, Includes the following steps: S1. Collect the sequence of answer events of the target student in the target question, wherein the sequence of answer events includes the operation type and the corresponding timestamp; S2. Based on the time interval between adjacent response events and the submission event, jump event, and rollback event, the response event sequence is segmented into multiple response step segments arranged in chronological order, and the step segment features are statistically analyzed for each step segment. The step segment features include at least the duration, number of operations, and number of rollbacks. S3. Obtain the standard problem-solving process corresponding to the target problem. The standard problem-solving process consists of multiple standard steps arranged in sequence and their corresponding key operation sets. S4. For each answer step segment, calculate its alignment matching degree with each standard step based on the features of the step segment and the key operation set, and determine the step alignment sequence under the condition of satisfying the standard step order constraint. At the same time, identify the unaligned standard steps, the repeatedly aligned answer step segments, and the alignment relationships with inconsistent order. S5. Based on the step alignment sequence of step S4 and the duration and number of operations of each response step segment, extract abnormal evidence units that represent missing steps, sequence shifts, repeated modifications and abnormal time consumption. S6. Match the abnormal evidence unit with the preset learning problem type determination rule, and output a structured learning problem localization result containing the problem type, corresponding standard steps and triggering abnormal evidence; S7. Based on the structured learning problem localization results, retrieve and output intervention element data from the intervention mapping table, wherein the intervention mapping table uses problem type and corresponding standard step identifier as index items.

2. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: Each answer event further includes an event target identifier, which includes at least one of the question stem area, option area, draft area, and formula area; and the operation type includes at least two or more of the following: input, deletion, option switching, formula editing, submission, withdrawal, rollback, and question jump.

3. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: The process segmentation in step S2 uses at least one of the following segmentation boundaries as the segment end boundary: (1) The time interval between adjacent response events is greater than the first time threshold; (2) A commit event occurs; (3) A question jump event occurs; (4) The number of backtracking events reaches the first threshold within the preset sliding time window; Furthermore, the first response event after the end boundary of the segment is used as the starting event for the next response step segment.

4. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: The step fragment features further include at least one or more of the following features: (1) Count vectors of different event types within a segment; (2) The number of deletions and the number of undoes within a segment shall be considered as the number of modifications; (3) Intra-fragment backspan, whereby the backspan is the number of times the editing position is bounced back or the number of steps the interface view is rolled back due to the backspan event; (4) Net time between the first input event and the first submission event within the segment.

5. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: Each standard step of the standard problem-solving process is further configured with a step attribute identifier, which includes at least one of key step identifiers and knowledge point identifiers. The set of key operations includes at least one of the following: key intermediate quantity formation, key formula calling, key condition judgment, and key option elimination.

6. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: Step S4, which determines the alignment sequence, includes a two-level determination process: The first determination process is as follows: For each response step segment, the matching degree is calculated based on the matching of the event type count vector of the response step segment with the key operation set to obtain a candidate alignment set; The second determination process: In the candidate alignment set, based on the order constraints of the standard step sequence, select the alignment combination that makes the overall alignment sequence consistent and maximizes the total matching degree; When a standard step is not aligned with any response step fragment, the standard step is marked as an unaligned standard step.

7. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: Step S5 generates a set of abnormal evidence units, which includes at least one or more of the following evidence units: E1. Evidence of missing steps: There are unaligned standard steps and the unaligned standard steps have critical step identifiers; E2, Sequence offset evidence: The number of inversion pairs of standard step identifiers in the step alignment sequence exceeds the second threshold; E3. Repeatedly modifying evidence: In the answer steps that are aligned with the same standard steps, delete the number of times the input is alternating with the deletion exceeds the third threshold, or the number of times the withdrawal is withdrawn exceeds the fourth threshold; E4. Evidence of abnormal time consumption: The duration of a response step segment aligned to the same standard step exceeds the fifth threshold of the reference duration of that standard step; E5. Evidence of backtracking: The cumulative backtracking span within a response step segment aligned to the same standard step exceeds the sixth threshold.

8. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 7, characterized in that: The reference duration is calculated from a reference sample set, which is a set of durations of step segments of multiple reference students on the corresponding standard steps under the same target question; the reference duration is taken as the median or quantile value of the duration set.

9. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: The structured learning problem localization result further includes evidence chain data, which includes at least the abnormal evidence unit identifier corresponding to the triggered problem type, the corresponding standard step identifier, and the start and end time range of the answer step segment corresponding to the abnormal evidence unit.

10. The data analysis method for personalized diagnosis and intervention of student learning problems according to claim 1, characterized in that: The intervention mapping table maps problem types and corresponding standard step identifiers to a set of intervention elements, and the set of intervention elements includes at least two of the following: intervention resource identifier, intervention task type, and intervention intensity level. The intervention task type includes at least one or more of the following: explanation task, error correction task, consolidation exercise task, and retest task. The intervention intensity level is determined by the number of abnormal evidence units triggered or the severity level of the evidence.

Images