Search

CN-122022153-A - Data analysis method for personalized diagnosis and intervention of student learning problem

CN122022153ACN 122022153 ACN122022153 ACN 122022153ACN-122022153-A

Abstract

The invention provides a data analysis method for personalized diagnosis and intervention of a student learning problem, which comprises the following steps of S1, collecting a response event sequence of a target student in a target question, wherein the response event sequence comprises an operation type and a corresponding time stamp, and S2, carrying out process segmentation on the response event sequence based on a time interval of adjacent response events, a submitting event, a jumping event and a rollback event. According to the invention, the process segmentation, the step alignment and the abnormal evidence extraction are carried out on the answer event sequence in the student answer process, the learning abnormality of the student is specifically positioned to the corresponding standard answer step and the trigger evidence thereof, and the structured learning problem positioning result is formed, so that the learning analysis result is converted into traceable and interpretable procedural diagnosis information from a single result index, the coarse granularity judgment is avoided depending on the answer result or the abnormal score, and the usability and pertinence of the learning problem analysis are improved.

Inventors

  • YANG BAI

Assignees

  • 爻象科技(广州)有限公司

Dates

Publication Date
20260512
Application Date
20260127

Claims (10)

  1. 1. The data analysis method for personalized diagnosis and intervention of the learning problem of the student is characterized by comprising the following steps: S1, collecting a response event sequence of a target student in a target question, wherein the response event sequence comprises an operation type and a corresponding time stamp; S2, carrying out process segmentation on the response event sequence based on the time interval of the adjacent response event, the submitting event, the jump event and the rollback event to obtain a plurality of response step fragments arranged in time sequence, and counting step fragment characteristics for each step fragment, wherein the step fragment characteristics at least comprise duration, operation times and rollback times; S3, acquiring a standard problem solving process corresponding to the target problem, wherein the standard problem solving process consists of a plurality of standard steps which are arranged in sequence and a key operation set corresponding to the standard steps; S4, aiming at each answer step segment, calculating the alignment matching degree between the answer step segment and each standard step based on the step segment characteristics and the key operation set, determining a step alignment sequence under the condition that the standard step sequence constraint is met, and simultaneously identifying an unaligned standard step, repeatedly aligned answer step segments and an alignment relation with inconsistent sequence; S5, extracting an abnormal evidence unit for representing step missing, sequence deviation, repeated modification and time consuming abnormality based on the step alignment sequence of the step S4 and the duration and operation times of each answering step segment; s6, matching the abnormal evidence unit with a preset learning problem type judgment rule, and outputting a structured learning problem positioning result comprising a problem type, a corresponding standard step and triggering abnormal evidence; And S7, retrieving and outputting intervention element data from an intervention mapping table according to the structured learning problem positioning result, wherein the intervention mapping table takes the problem type and the corresponding standard step identification as index items.
  2. 2. The method for personalized diagnosis and intervention data for a student's learning problem according to claim 1, wherein each response event further comprises an event action object identifier, wherein the event action object identifier comprises at least one of a stem area, an option area, a draft area and a formula area, and the operation type comprises at least two or more of input, deletion, option switching, formula editing, submission, withdrawal, rollback and question jumping.
  3. 3. The method for personalized diagnosis and intervention of learning problems of students according to claim 1, wherein the process segmentation in the step S2 adopts at least one of the following segmentation boundaries as a segment ending boundary: (1) The adjacent answer event time interval is greater than a first time threshold; (2) A commit event occurs; (3) A topic jump event occurs; (4) The number of rollback events reaches a first time threshold within a preset sliding time window; And the first answer event after the end boundary of the segment is taken as the start event of the segment of the next answer step.
  4. 4. The method for personalized diagnosis and intervention of a learning problem of a student according to claim 1, wherein the step segment features further comprise at least one or more of the following features: (1) Count vectors for different event types within a segment; (2) At least one of the number of intra-segment deletions and the number of withdrawals is used as a modification number; (3) The method comprises the steps of intra-segment backspacing spans, wherein the backspacing spans are editing position rebound times or interface view backspacing numbers caused by backspacing events; (4) The net time between the first input event to the first commit event within the segment.
  5. 5. The method for personalized diagnosis and intervention data analysis of student learning problems according to claim 1, wherein each standard step of the standard solution process is further configured with a step attribute identifier, and the step attribute identifier at least comprises at least one of a key step identifier and a knowledge point identifier; the key operation set at least comprises at least one of key intermediate quantity formation, key formula call, key condition judgment and key option exclusion.
  6. 6. The method for personalized diagnosis and intervention of learning problems for students according to claim 1, wherein the step S4 of determining the step alignment sequence comprises two steps of determining: The first determining process is to calculate the matching degree of each answer step segment according to the matching condition of the event type count vector of the answer step segment and the key operation set to obtain a candidate alignment set; The second determining process is to select an alignment combination with the largest sum of matching degrees and the whole sequence of the alignment sequences being consistent according to the sequence constraint of the standard step sequence in the candidate alignment set; When a standard step is not aligned by any of the answer step fragments, the standard step is marked as a misaligned standard step.
  7. 7. The method for personalized diagnosis and intervention of a learning problem of a student according to claim 1, wherein the step S5 of generating a set of abnormal evidence units comprises at least one or more of the following evidence units: E1, step missing evidence that an unaligned standard step exists and has a key step identifier; e2, sequentially shifting evidence that the number of the reverse sequence numbers exceeds a second threshold value when the standard step marks appear in the step alignment sequence; e3, repeatedly modifying evidence, namely deleting and inputting the pieces of answering steps with the alternation times exceeding a third threshold value or withdrawing the pieces of evidence exceeding a fourth threshold value in the pieces of answering steps aligned to the same standard step; e4, time-consuming abnormal evidence that the duration of the answering step fragments aligned to the same standard step exceeds the fifth threshold multiple of the reference duration of the standard step; and E5, backing off loiter evidence that the cumulative value of the backing off span in the answer step segment aligned to the same standard step exceeds a sixth threshold.
  8. 8. The method for personalized diagnosis and intervention data analysis of student learning problems according to claim 7, wherein the reference time length is calculated by a reference sample set, the reference sample set is a step segment duration set of a plurality of reference students on corresponding standard steps under the same target subject, and the reference time length takes a median or a quantile value of the duration set.
  9. 9. The method for personalized diagnosis and intervention of a student learning problem according to claim 1, wherein the structured learning problem localization result further comprises evidence chain data, wherein the evidence chain data at least comprises an abnormal evidence unit identifier corresponding to a triggered problem type, a corresponding standard step identifier, and a starting and ending time range of a answering step segment corresponding to the abnormal evidence unit.
  10. 10. The method for personalized diagnosis and intervention data analysis of student learning problems according to claim 1, wherein the intervention mapping table maps the problem types and the corresponding standard step identifiers into an intervention element set, and the intervention element set at least comprises at least two of intervention resource identifiers, intervention task types and intervention intensity levels; the intervention task type at least comprises one or more of explanation tasks, error correction tasks, consolidation training tasks and retesting tasks, and the intervention intensity level is determined by the number of triggered abnormal evidence units or the evidence severity level.

Description

Data analysis method for personalized diagnosis and intervention of student learning problem Technical Field The invention relates to the technical field of learning data analysis, in particular to a data analysis method for personalized diagnosis and intervention of student learning problems. Background With the popularity of online learning platforms, intelligent classes, and homework systems, the teaching process generates a large amount of data related to student learning activities, including but not limited to homework and quiz records, time and number of remakes for answering questions, learning resource clicks and viewing trajectories, class interaction text, and structured log data related to the learning process. Based on the data, the learning state analysis is developed, and becomes an important technical direction of teaching management and learning support; In practical teaching application, a teacher or a learning support system generally needs to perform targeted intervention based on an explicit learning problem positioning result, such as performing explanation around a specific knowledge point, performing error correction training for a specific error type or pushing corresponding learning resources, if an analysis result lacks interpretable diagnosis information, only abnormal scores are provided, and subsequent personalized intervention decisions are difficult to directly support, and still manual analysis is needed to influence the efficiency and accuracy of teaching support; Therefore, a data analysis method for personalized diagnosis and intervention of student learning problems is provided. Disclosure of Invention In view of the above, the present invention provides a data analysis method for personalized diagnosis and intervention of learning problems of students, so as to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial choice. The technical scheme of the invention is realized in such a way that the data analysis method for personalized diagnosis and intervention of the learning problem of the student comprises the following steps: S1, collecting a response event sequence of a target student in a target question, wherein the response event sequence comprises an operation type and a corresponding time stamp; S2, carrying out process segmentation on the response event sequence based on the time interval of the adjacent response event, the submitting event, the jump event and the rollback event to obtain a plurality of response step fragments arranged in time sequence, and counting step fragment characteristics for each step fragment, wherein the step fragment characteristics at least comprise duration, operation times and rollback times; S3, acquiring a standard problem solving process corresponding to the target problem, wherein the standard problem solving process consists of a plurality of standard steps which are arranged in sequence and a key operation set corresponding to the standard steps; S4, aiming at each answer step segment, calculating the alignment matching degree between the answer step segment and each standard step based on the step segment characteristics and the key operation set, determining a step alignment sequence under the condition that the standard step sequence constraint is met, and simultaneously identifying an unaligned standard step, repeatedly aligned answer step segments and an alignment relation with inconsistent sequence; S5, extracting an abnormal evidence unit for representing step missing, sequence deviation, repeated modification and time consuming abnormality based on the step alignment sequence of the step S4 and the duration and operation times of each answering step segment; s6, matching the abnormal evidence unit with a preset learning problem type judgment rule, and outputting a structured learning problem positioning result comprising a problem type, a corresponding standard step and triggering abnormal evidence; And S7, retrieving and outputting intervention element data from an intervention mapping table according to the structured learning problem positioning result, wherein the intervention mapping table takes the problem type and the corresponding standard step identification as index items. Further preferably, each answer event further comprises an event action object identifier, wherein the event action object identifier at least comprises one of a question stem area, an option area, a draft area and a formula area, and the operation type at least comprises two or more of input, deletion, option switching, formula editing, submission, withdrawal, rollback and question jumping. Further preferably, the process segmentation in the step S2 uses at least one of the following segmentation boundaries as a segment end boundary: (1) The adjacent answer event time interval is greater than a first time threshold; (2) A commit event occurs; (3) A topic jump event occurs; (4) The number of