Adaptive teaching system based on learning behavior analysis

Abstract

The invention relates to the technical field of reasoning models, in particular to an adaptive teaching system based on learning behavior analysis, which comprises a learning focusing behavior description module, an operation section interference identification module, a behavior offset section identification module, a teaching node jump path construction module and a teaching task configuration generation module. According to the method, time sequences of operation behaviors are divided, path change features are extracted, a behavior focusing fragment and interference node set is constructed, a path fracture position is identified by combining a jump direction and an offset action, and a jump path is generated according to a label repetition frequency and a non-present node matching relation. Task styles, resource presentation modes and interaction process contents are extracted, a task distribution structure between nodes is constructed, the path structure reduction capability and the task content cooperation capability are enhanced, and smooth task node connection and coherent content pushing are guaranteed under the scenes of frequent skipping of learning behaviors and disordered path sequence; and the coordination efficiency of the teaching process on the scheduling response of the behavior change is improved.

Description

Technical Field

[0001] This invention relates to the field of reasoning model technology, and in particular to an adaptive teaching system based on learning behavior analysis. Background Technology

[0002] The field of reasoning model technology involves a methodological technical system for logical judgment and result deduction based on input data using a predetermined knowledge representation. The core aspects of this technology include determining the knowledge representation, constructing reasoning rules, designing the reasoning process, and completing the deduction judgment based on the relationship between input data and existing knowledge. It typically uses rule sets, probability relationships, or behavioral patterns as the basis for reasoning, and forms the reasoning result through matching, comparison, and logical operations on the input information. This type of method is widely used in computer systems that require judgment and decision-making based on historical data or behavioral characteristics. Traditional adaptive teaching systems based on learning behavior analysis refer to teaching systems that adjust teaching content and sequence based on students' behavioral records during the learning process. Traditional methods for determining teaching content configuration and teaching process arrangement based on learning behavior data typically involve collecting specific behavioral data such as students' answers, learning time, page visits, and learning progress records. This behavioral data is then compared with pre-set teaching rules or thresholds to determine the selection order and difficulty setting of subsequent teaching content. The relevant process mainly relies on fixed rule tables, behavioral statistical results, and preset judgment conditions to complete the matching and adjustment of teaching content.

[0003] Existing technologies rely on preset rules and fixed conditions to statically judge the results of learning behaviors. When dealing with complex and intertwined operational behaviors and interrupted path execution, they lack the ability to process segments and the mechanism to restore the structure. The behavioral data only presents limited index values ​​and distribution characteristics, which are difficult to reflect the change patterns of broken areas in the operation chain and the logical transformation relationship between jump paths. In the case of frequent question type switching and misaligned operation rhythm, the system cannot effectively track the trend of behavioral changes and the gaps in path connection, which may lead to the disconnection of content push and the misalignment of task flow, weakening the coherent response capability of the teaching process and the accuracy of behavior mapping. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide an adaptive teaching system based on learning behavior analysis.

[0005] On the one hand, an adaptive teaching system based on learning behavior analysis is provided, which includes: The learning focus module obtains the page dwell order, click jump order and operation trajectory, tracks page switching and position changes, divides the operation sequence into segments, and obtains a list of operation focus behavior segments; The operation segment interference identification module, based on the operation focused behavior segment list, compares the question type structure with the operation sequence, tracks the changes in actions before and after the jump behavior, and obtains a set of behavior interference segment identifiers. The behavior offset segment identification module extracts jump behavior, misdirected click and back point based on the behavior interference segment identifier set, compares the direction change and action repetition, and obtains the path structure reconstruction segment set according to the order between the task abandonment position and the jump start position. The teaching node jump path construction module reconstructs the fragment set based on the path structure, extracts duplicate label nodes, finds label nodes that do not appear in the path, and advances the jump path according to the time sequence of the operation behavior to obtain the teaching jump node sequence group. The teaching task configuration generation module extracts task styles, resource methods and interaction flows based on the teaching jump node sequence group, compares the task content of adjacent nodes, analyzes the distribution of tasks in the path, and obtains a set of teaching task execution methods.

[0006] As a further aspect of the present invention, the list of operation-focused behavior segments includes page dwell order, click jump order, operation path trajectory, page switching frequency, click position repetition, and time sequence arrangement. The set of behavior interference segment identifiers includes jump behavior position distribution, question type structure differences, operation sequence interruption points, page jump direction changes, and operation chain interruption nodes. The set of path structure reconstruction fragments includes jump behavior sequences, misdirected click actions, page back points, click direction change order, click target inconsistency segments, and the mapping relationship between task abandonment position and jump start position. The teaching jump node sequence group includes repeated label nodes, non-appearing node set, label-covered content items, label name correspondence, and time sequence extension nodes. The set of teaching task execution methods includes task style type, resource presentation form, interaction flow structure, question type structure mapping method, content direction change characteristics, and task style distribution.

[0007] As a further aspect of the present invention, the question type structure refers to the question type and structural features bound to each learning page or node; The task abandonment location refers to the node and page where a student leaves without completing a learning task due to interruption, exit, or incorrect clicks.

[0008] As a further aspect of the present invention, the jump start position refers to the starting point from which the student begins to perform the jump operation from the page; The resource method refers to the presentation format of the teaching content on the page.

[0009] As a further aspect of the present invention, the learning-focused behavior characterization module includes: The page behavior sequence acquisition submodule acquires the order of student's page stay, click jump order and operation path trajectory in the question reading area, option browsing area and submission operation area. Based on the time sequence of page stay actions and click actions, it extracts the action sequence of page switching and position change, analyzes the order of actions on the timeline, and obtains the page operation behavior sequence. The operation trajectory segmentation submodule, based on the page operation behavior sequence, extracts the behavior trajectory where the page does not switch but the click position is continuously repeated according to the correspondence between the page dwell position and the time of the click action, and draws out the operation segments where the page does not change continuously, thus obtaining the operation trajectory segment set; The focus segment extraction submodule extracts segments with continuous page switching based on the operation trajectory segment set, and obtains a list of operation focus behavior segments by corresponding to the page number and click position number in each segment.

[0010] As a further aspect of the present invention, the operation segment interference identification module includes: The path behavior structure comparison submodule extracts the question type structure number, page dwell order and operation click order number in the previous and subsequent paths based on the operation focus behavior segment list, and compares the positions in the path where the order of appearance of question type structure and operation order change to obtain the question type behavior interruption interval set. The page jump direction analysis submodule extracts the starting page and target page direction numbers based on the question type behavior interruption interval set, compares the page jump direction changes in the path segment, identifies operation segments where the jump direction is repeatedly switched between adjacent positions, and obtains the page jump direction change segment set. The interference node extraction submodule extracts the click node numbers between the first time the page jumps out and the last time the page is returned during the jump operation based on the set of page jump direction change segments. The click nodes within the segments are then aggregated to obtain a set of behavior interference segment identifiers.

[0011] As a further aspect of the present invention, the behavior offset segment identification module includes: The behavior action sequence extraction submodule extracts jump behaviors, misdirected click actions, and page back points within the path segment based on the behavior interference segment identifier set. It then identifies repeated behaviors by corresponding page numbers, click positions, and direction changes in time sequence to obtain a set of operation behavior feature sequences. The path direction change recognition submodule calls the operation behavior feature sequence group to analyze the correspondence between the direction number and the click position change during the page jump process, identifies the position where the forward and backward directions change in the path segment, extracts the click action where the direction is inconsistent with the target position, and obtains the direction alternation behavior index set; The offset segment corresponding mapping submodule extracts the clicked nodes between the task abandonment position and the jump start position based on the direction alternation behavior index set, compares the operation corresponding to the node in the segment with the original action sequence in the path, and obtains the path structure reconstruction fragment set.

[0012] As a further aspect of the present invention, in the process of extracting jump behavior, misdirected click action and page back point within the path segment: within the page range covered by the same behavior interference segment identifier set, identify click actions with consecutive time intervals, classify the jump behavior to which the consecutive clicks belong, and extract click actions with reverse direction offset according to the spatial change trend between the click position and the page jump direction. In the process of identifying recurring behaviors: in the group of operation behavior feature sequences, the correspondence between page number and click position is compared, and the same operation behavior features that recur in the position sequence are filtered. During the process of extracting the click nodes between the task abandonment position and the jump start position: based on the jump start point shown in the direction alternation behavior index set, trace forward along the user operation sequence and extract the click nodes in the path after the jump start point.

[0013] As a further aspect of the present invention, the teaching node jump path construction module includes: The path segment extraction submodule reconstructs a segment set based on the path structure, extracts the node numbers and content tags corresponding to the operation behaviors in the path, compares the number of times the node numbers are repeated, removes the numbers whose number of repetitions exceeds a preset frequency threshold, and obtains the path behavior time series node numbers. The tag matching and recognition submodule calls the node tags in the path behavior time series node numbers, matches each item in the content tag set with the behavior tag, identifies the node tag set that does not appear in the path, filters the tag items with the same number of repetitions, and locates the corresponding node number to obtain the node number set with the same number of tag repetitions; The jump node sequence output submodule compares the node numbers in the set of node numbers with the same number of repetitions of the tags with the range of the number of tags corresponding to the content tags, selects the content item numbers with the same number of tags, and extends the node number sequence according to the time order of the path behavior time sequence node numbers to obtain the teaching jump node sequence group.

[0014] As a further aspect of the present invention, the teaching task configuration generation module includes: The task parameter extraction submodule extracts the task style, resource presentation method and interaction flow associated with the node based on the teaching jump node sequence group. It extracts the resource structure corresponding to the task style according to the node order, corresponding resource number and task style identifier, and obtains the node task parameter merge list. The operation style mapping submodule extracts the sequence of executed operation methods based on the task styles and interaction flow content in the node task parameter merge list, compares the question structure and material format style, and marks the related operation styles to the corresponding nodes to obtain the node operation style mapping structure. The execution method analysis submodule compares the content direction and task style differences between adjacent nodes based on the node operation style mapping structure, groups them into stages according to the jump path, detects the distribution of task styles in each group, and obtains the set of teaching task execution methods.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by dividing the time series of operational behaviors and extracting path change features, a set of behavior-focused segments and interference nodes is constructed. The path breakage position is identified by combining the jump direction and offset action. Jump paths are generated based on the repetition frequency of tags and the matching relationship with absent nodes. Task styles, resource presentation methods and interactive flow content are extracted, and a task distribution structure between nodes is constructed. This enhances the ability to restore the path structure and the ability to coordinate task content. In scenarios where learning behaviors frequently jump and the path order is disordered, it ensures smooth connection of task nodes and coherent content push, and improves the coordination efficiency of the teaching process in responding to behavioral changes. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a system block diagram of the present invention; Figure 3 This is a flowchart of the learning-focused behavior characterization module in this invention; Figure 4 This is a flowchart of the operation segment interference identification module in this invention; Figure 5 This is a flowchart of the behavior offset segment identification module in this invention; Figure 6 This is a flowchart of the teaching node jump path construction module in this invention; Figure 7 This is a flowchart of the teaching task configuration generation module in this invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0020] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0021] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0022] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0023] All user-related information involved in this invention (including but not limited to biometric information, identity verification information, behavioral data, device information, and other data that can be used for identity verification and personalized services) is collected and processed with the user's full knowledge and voluntary consent. The collection, storage, and use of all information strictly comply with applicable national and regional laws and regulations, and meet relevant data protection standards and policy requirements. The use of data is limited to purposes necessary for providing the technical services of this invention, and reasonable technical and management measures will be taken to ensure the security and confidentiality of users' personal information in terms of information protection and privacy.

[0024] This invention provides an adaptive teaching system based on learning behavior analysis, such as... Figure 1-2 The diagram shown illustrates an adaptive teaching system based on learning behavior analysis. The system includes: The learning-focused behavior characterization module obtains the order of students' page dwell time, click jump order, and operation path trajectory in the question stem reading area, option browsing area, and submission operation area. It tracks the arrangement of page switching actions and operation position changes in the time series, divides the page dwell time and click action into segments according to the order of occurrence, identifies operation processes with a limited number of page switching times and repeated click positions, and obtains a list of operation-focused behavior segments. The operation segment interference identification module is based on the operation focused behavior segment list. It compares the question structure, operation order and jump behavior position in the previous and subsequent paths, tracks the interruption position between consecutive question types in the operation chain and the change of page jump direction, locates the position where the action order in the path is interrupted, and extracts the operation node corresponding to the interruption area according to the distribution relationship of the jump behavior position in the path segment to obtain the behavior interference segment identifier set. The behavior offset segment identification module extracts consecutive jump behaviors, misdirected click actions, and page back points in the interfered path segment based on the behavior interference segment identifier set. By comparing the change order of click direction and the number of repetitions of operation actions within the path, it identifies the position segments in the path where the direction alternates and the click target is inconsistent. Then, it matches the operation order between the task abandonment position and the jump start position to obtain the path structure reconstruction segment set. The teaching node jump path construction module reconstructs the fragment set based on the path structure, extracts nodes from the content tag set that have the same number of repetitions as the current behavior fragment tag, searches for nodes that do not appear in the current path, extracts content items with a fixed number of tags from the learning progress, matches the tag names with the existing tags in the missing nodes, and extends the operation nodes backward according to the time sequence of the operation behavior in the original path to obtain the teaching jump node sequence group. The teaching task configuration generation module is based on the teaching jump node sequence group. It extracts the corresponding task style, resource presentation method and interaction flow in the node, maps the question structure, material form and the operation method performed by the user, compares the content direction and task style between adjacent nodes, analyzes the distribution of task style in the jump path, and obtains the set of teaching task execution methods.

[0025] The list of operation-focused behavior segments includes page dwell order, click jump order, operation path trajectory, page switching frequency, click position repetition, and time sequence arrangement. The set of behavior interference segment identifiers includes the distribution of jump behavior positions, differences in question type structure, operation sequence interruption points, page jump direction changes, and operation chain interruption nodes. The set of path structure reconstruction fragments includes jump behavior sequences, misdirected click actions, page back points, click direction change order, click target inconsistency segments, and the mapping relationship between task abandonment position and jump start position. The teaching jump node sequence group includes repeated label nodes, non-appearing node sets, label-covered content items, label name correspondence, and time sequence extension nodes. The set of teaching task execution methods includes task style type, resource presentation form, interaction flow structure, question type structure mapping method, content direction change characteristics, and task style distribution.

[0026] Specifically, such as Figure 2 , 3 As shown, the learning-focused behavioral characterization module includes: The page behavior sequence acquisition submodule acquires the order of student's page stay, click jump order and operation path trajectory in the question reading area, option browsing area and submission operation area. Based on the time sequence of page stay actions and click actions, it extracts the action sequence of page switching and position change, analyzes the order of actions on the timeline, and obtains the page operation behavior sequence. Before acquiring student action data, the system first obtains objective factors such as the student's current preset learning goals, real-time learning status, and initial cognitive abilities assessed based on historical data. These serve as a benchmark for subsequently eliminating behavioral ambiguity and accurately determining cognitive needs. The system acquires the student's page dwell order, click sequence, and action path trajectory in the question stem reading area, option browsing area, and submission area. The page dwell order can be determined by recording the initial entry and exit sequence numbers for each area. For example, if the student enters the question stem reading area, then the option browsing area, and then returns to the question stem reading area, this sequence should be marked as 1-2-1. Simple behavioral sequences can be ambiguous, requiring judgment in conjunction with the student's objective factors: for example, if the system records that the student has weak cognitive abilities and has not mastered the relevant concepts, this 1-2-1 sequence typically reflects an inability to establish a connection between the question stem and the options, indicating cognitive difficulty; if the student has strong cognitive abilities and is in a review state, this sequence is judged as normal comparative thinking behavior. By combining objective factors and behavioral sequences, obtaining the click-to-jump order requires extracting the click node numbers of element-triggered events. The page numbers are then serialized according to the order in which the click events are triggered. Simultaneously, non-jump click events, such as option highlighting or hover tooltip events, must be excluded to avoid interfering with the page jump path. Obtaining the operation path trajectory requires calling the event chain in the client's behavior log, extracting the page numbers between the click start and end points, and using each page stay and jump as a boundary to continuously segment the path. For example, if the operation trajectory is "Click on page A—Browse page B—Click on page C," then the path should be recorded in two segments in the order A→B and B→C. Next, based on the order of page stay actions and click actions, the associated nodes between the end of each stay and the triggering of the click action are identified. The page identifier at the end of the stay action is paired with the page identifier after the jump. For example, if a stay occurs at... If a click occurs after page A, leading to page C, this sequence should be marked as A→C. Each pause and jump action is concatenated according to the original order recorded in the client logs to form a complete set of user operation path segments. All page numbers and location numbers in the above sequence are compared one-to-one to determine if the click action caused a change in location number. If the location number does not change in two consecutive records, it is marked as a jump within the same segment; otherwise, it is marked as a path node switch. Based on the jump frequency within the sequence and the degree of difference between the start and end page numbers, it is determined whether the page jump is a page content switch or a function switch within the same area. For example, if the page numbers are consecutively A→A→B, and the location numbers are different in the first two records, it is considered a function switch within the same area. If the page numbers change to A→B→C, and the location numbers belong to different segments, it is considered a page level switch. Following the above processing method, the data combination process for all operation trajectory segments is executed sequentially to finally obtain the page operation behavior sequence.

[0027] The operation trajectory segmentation submodule is based on the page operation behavior sequence. According to the correspondence between the page dwell position and the time of the click action, it extracts the behavior trajectory where the page does not change but the click position is continuously repeated, and draws out the operation segments where the page does not change continuously, thus obtaining the operation trajectory segment set. First, extract the continuity feature of the page number and click position number combination in each behavior sequence. Compare each pair of consecutive page numbers in the sequence. If the page numbers of two consecutive pages are consistent, it is determined that the position has not changed. Next, the click event number under the corresponding position needs to be retrieved, and the consecutive click positions are numbered and recorded. If multiple click position numbers appear consecutively on the same page and the position numbers are the same, it is marked as consecutive clicks. For example, if click actions with the numbers P1, P1, and P1 occur consecutively on page A, and the page number is always A, then this behavior segment should be extracted as a consecutive click position behavior trajectory segment. Conversely, if the page number changes after any click, the segment is stopped and considered as a behavior interruption point. For the above operation sequence, it is necessary to further determine whether the jump between the start and end positions of the action before and after the page switch is accompanied by a change in the page number. For example, if the page sequence is AAABCC and the click position numbers are P1-P1-P2-P3-P4-P4, it can be divided into consecutive P1 within segment A, page change after P2 appears within segment A, etc. There is no click continuity within segments B and C. In this sequence, only the P1 behavior in segment A meets the continuous click trajectory condition. During processing, it is also necessary to exclude page number changes caused by non-user-triggered interface refreshes or system prompts. Fields marked as "automatic jump" or "background refresh" in the operation record are removed to ensure that the extraction results only contain user behavior paths. Then, based on the effective behavior segments extracted above, the starting page number and ending page number of each segment are extracted, and it is checked whether the page has changed. If the page number has not changed, the segment is written into the trajectory set as a stable page operation record. At the same time, the change of click position number is recorded. For example, if a segment is page B and the click position numbers are P2, P2, P3, P2, it is determined that there is position repetition in the segment but the page remains unchanged, which meets the continuous behavior requirement. The content of this segment is classified into the identifiable trajectory segment. After traversing all page operation behavior sequences, the page number and click position number recorded in each segment form a path group, which is uniformly identified and output to obtain the operation trajectory segment set.

[0028] The focus segment extraction submodule extracts segments with continuous page switching based on the operation trajectory segment set, and obtains a list of operation focus behavior segments by corresponding to the page number and click position number in each segment. First, extract the page number sequence and corresponding click position number sequence from each segment in sequence. For the page number sequence, compare the page identifier between every two adjacent numbers to see if they change. If the page number remains consistent between two positions, it is recorded as a state where the page has not jumped. If three consecutive page numbers in a segment are A, and the click positions are P1, P1, and P2 respectively, then the first two actions in this segment constitute a repeated click behavior, while the third action, although on the same page, is at a different click position and should be considered a position jump within the same area. For example, in another segment, the page numbers are A, A, A, and the click position numbers are P2, P2, and P2. Then the entire segment consists of repeated clicks and the page remains unchanged. This segment can be determined as a trajectory segment where the number of page switches remains continuous. For the processing of click position numbers, the position code field in the click event record needs to be read sequentially and matched character-level with the previous record. If the two codes are completely identical, it is marked as a repeated position. If the position codes only have the same prefix but different suffixes, it is determined as a repeated position. For click offset actions within the same block, it is necessary to compare them with the page number to determine whether they are repeated operations within the same view area. Further, the combination relationship between page number and position number in such segments is extracted to form a sequence set of page-position pairs. These pairs are then associated by segment number, and record groups where the page has not changed and the click actions occur consecutively are filtered out. Each of these record groups contains at least three consecutive actions. If a group contains only two clicks at the same position but the interval exceeds the average click interval within the page, it is not counted as a valid segment. The average click interval within the page can be calculated based on the consecutive click time difference in the historical behavior sequence of that page. Taking page A as an example, if the average interval between consecutive clicks at position P1 on page A in all past trajectories is 2 seconds, while the interval between two clicks at position P1 in the current behavior is 5 seconds, then the group should be excluded. Otherwise, it should be retained. After completing the above processing, all segment numbers that meet the criteria of unchanged page number, repeated click positions, and reasonable operation intervals are written into the result list to obtain the operation-focused behavior segment list.

[0029] Specifically, such as Figure 2 , 4 As shown, the operation segment interference identification module includes: The path behavior structure comparison submodule extracts the question type structure number, page dwell order and operation click order number in the previous and subsequent paths based on the list of operation focused behavior segments. It compares the positions in the path where the order of appearance of question type structure and operation order change to obtain the question type behavior interruption interval set. First, extract the question type structure number sequence involved in the paths before and after each segment. Then, compare the page numbers and operation click numbers within each path segment. First, based on each page number in the path, extract the number by calling its associated question type structure identifier, forming a question type structure number sequence. Next, extract the page entry order based on the dwell order field in the page records. This field is usually arranged according to the entry time and can be directly represented by the number field. For example, if page B enters after page A, the numbers are A1 and B2. Then, sort the click action number sequence, forming an operation click number sequence based on the triggering order of each click event in the records. This sequence is numbered sequentially by trigger time, such as click positions P1, P3, etc. If P2 appears sequentially, the click sequence is P1-1, P3-2, P2-3. After extracting the question structure number, page sequence number, and click sequence number, the order of the three types of numbers needs to be compared to determine if there is a discrepancy between the order of appearance of the question structure number and the order of operation clicks. For example, if the question structure sequence is T1, T2, T3, and the corresponding click sequence is T1-P1, T3-P2, T2-P3, then it can be identified that the click order of T2 and T3 has been reversed. Further comparison is made between the page order and the question structure number order. If the page number bound to the question structure T3 appears after T2, but the click action occurs before T2, it means that the user skipped the content of T2. At this point, it is necessary to explore the reasons for skipping based on the objective factors such as the student's cognitive ability and learning status obtained above: if the student's objective ability data shows that they have mastered the knowledge points corresponding to T2, then this reverse order operation is a strategic way of doing questions (starting with the difficult ones and then moving to the easy ones); if the student has not mastered T2 and has weak cognitive ability, then this skipping represents cognitive abandonment behavior due to encountering knowledge difficulties. Based on this logic, such sequence misalignment cases need to be recorded according to the path number range. All positions where the sequence changes occur are written into the record table to form a number correspondence table. Then, the number boundaries in the path behavior segments are combined to define the segments. Segments with adjacent start and end numbers but misaligned sequences are written into the structure as identifiable fragments. For example, if the path behavior numbers are A, B, C, and D, the question type structure is T1, T2, T3, and T4, and the click order is T1, T3, T2, and T4, then the misalignment interval formed between T2 and T3 can be extracted. At the same time, it is determined whether the misalignment occurs continuously. If multiple consecutive question type structures have this kind of reverse operation, they are merged into a longer interruption interval. Finally, all number pairs where the question type structure and operation order are inconsistent are output and grouped according to the path segment order to obtain the question type behavior interruption interval set.

[0030] The page jump direction analysis submodule extracts the starting page and target page direction numbers based on the question type behavior interruption interval set, compares the changes in page jump direction in the path segment, identifies the operation segment where the jump direction is repeatedly switched between adjacent positions, and obtains the page jump direction change segment set; First, the path segment number range corresponding to each interruption interval is read, and the starting page number and target page number in the page jump record are extracted sequentially within this range. Each jump is regarded as a direction event, and the direction event is represented by the starting page number pointing to the target page number. For example, a jump from page A to page B is recorded as A to B. Then, consecutive direction events within the same interval are read sequentially to form a page jump direction sequence. During the processing, the interval boundary needs to be limited, and only jump records within the same interruption interval number range are processed to avoid cross-segment data interference. Next, the direction relationship between two adjacent jumps in the direction sequence is compared item by item to determine whether the starting page and the target page have reversed or alternated. For example, if the consecutive direction events are A to B, B to A, and A to B, then the direction changes repeatedly in this sequence. In the educational logic of cognitive load theory, frequent reversals of direction between adjacent pages indicate that students' working memory is overloaded, making it impossible to effectively process and integrate information in a single reading or single-page stay. They must rely on repeated references to compensate for insufficient memory retention. Therefore, this specific sequence characteristic objectively defines a student's "cognitive impairment state." For example, if the directional event is C to D, C to D, C to D, the direction remains consistent and is not recorded. During the comparison, the page stay records need to be read simultaneously to confirm whether the page stay number before each jump is consistent. If the page stay number is the same but the direction changes, it is determined to be a directional change within the same page. If the page stay number is different but the direction changes, it is determined to be a cross-page directional change. Both types of cases need to be recorded separately. Subsequently, all position numbers where directional changes occur are marked in the directional sequence, indicating continuous directional changes. The positions are merged into one operation segment. For example, if the 3rd to 6th jumps in a certain interval all show alternating directions, then this numbered range should be recorded as a complete segment. Further, the number of direction changes within this segment is counted. The number of direction changes is considered as one change when two adjacent directions are inconsistent. For example, if the direction sequence A to B, B to A, and A to B contains 2 changes, this value needs to be recorded together with the segment length as the basis for subsequent segment determination. In this process, non-question-related jump events such as page refresh or returning to the homepage need to be excluded. They can be distinguished based on the operation source field in the jump record. Only the page jump records actively triggered by the user are retained. After completing the above processing, the same direction reading and comparison process is performed on all interrupted intervals. All the numbered segments with repeated direction changes are written into the result list in path order to obtain the set of page jump direction change segments.

[0031] The interference node extraction submodule extracts the click node numbers between the first time the page jumps out and the last time the page is returned during the jump operation based on the set of segments where the page jump direction changes. The click nodes are then matched with the operation actions in the order of the numbers, and the click nodes within the segments are aggregated to obtain a set of behavior interference segment identifiers. First, extract the starting page number from the first page jump record in this segment, and mark it as the jump start point. Continue searching downwards until a jump operation with a target page number equal to the starting page number is found, and mark it as the last return page. All records between these two jump operations are the valid processing interval. Then, extract the click event records within this interval. The node number in the click record can be obtained through the click target field in the event sequence. This field usually corresponds to the element identifier on the page that is bound to the click event. For example, node number N1 indicates clicking the "Next Page" button, node number N2 indicates clicking the "Back" icon, and so on. When extracting node numbers, it is necessary to remove non-interactive nodes, such as static display areas, whose clicks will not trigger page jump logic. It is necessary to filter by the event type field of the click behavior, and only retain click behaviors that belong to active interactive operations, such as click behaviors of type "jump trigger" or "view switching". After extraction, map the node numbers according to their order in the event records to construct the interval. The sequence of click actions within a given timeframe retains the page number corresponding to each click, allowing for subsequent tracking of the click node's position within the page navigation chain. For example, if the sequence is N1-N3-N2, corresponding to page A, page B, and page A again, then N3 is located on page B, indicating a cross-page click during the navigation process. Next, all node numbers in the sequence are extracted, and based on the correspondence between page numbers and node positions before and after the navigation, each click node is assigned a position, determining whether it appears on the transition page between the starting and target pages. If a node number only appears in the navigation path with the intermediate page number, it should be considered an interference node in the navigation process. These nodes need to be marked, and their position number in the event sequence should be associated with the click time order. Finally, the interference node numbers identified in each directional change segment and their operation order are summarized to form a unified number sequence set, recording its corresponding segment number and page navigation chain number, thus obtaining a behavioral interference segment identifier set.

[0032] Specifically, such as Figure 2 , 5 As shown, the behavior offset segment recognition module includes: The behavior action sequence extraction submodule extracts jump behaviors, misdirected click actions, and page back points within the path segment based on the behavior interference segment identifier set. It then identifies repeated behaviors by corresponding page numbers, click positions, and direction changes in time sequence to obtain a set of operation behavior feature sequences. First, within each identified segment, jump behavior records, incorrect direction click actions, and page back actions are extracted sequentially. The identification of jump behavior is based on the starting page number and target page number recorded in the page jump event. A page sequence is established according to the jump event order, and its corresponding jump start and end pairs are marked. For example, if the page numbers recorded in a certain path segment are A to B, B to C, C to B, and B to A in sequence, then it contains two forward jumps and two reverse jumps. The jump direction can be marked according to the difference in the page number sequence. If the target page number is less than the starting page number, it is considered a reverse jump, and otherwise it is a forward jump. To extract misdirected click actions, it is necessary to traverse the node numbers and jump targets recorded in the click event log, and determine whether the page jump direction triggered by it violates the normal process set by the system. If a node click operation causes a reverse jump and is not triggered by the page back button, it can be regarded as an misdirected click. For example, clicking the node with the number N5 causes the page to jump from D to B. If the page order set by the system is ABCD, then this jump constitutes an unexpected reverse jump. Identifying page back points requires judging the continuity of the jump direction. When multiple reverse jumps occur consecutively in the path sequence, such as the pattern of D to C, C to B, and B to A, in educational behavior, it usually indicates that the student is retrospectively searching for forgotten prior knowledge nodes. Information such as the start and end points of page jumps, the order of directional changes, and the source of jump triggers can be combined to identify whether typical pages in the intermediate process (such as page B) meet the explicit back characteristics, but this is not used as a direct identification basis. Finally, each jump, click, and possible back operation needs to be sorted by calling the timestamp field and marked with its position number in the behavior sequence to obtain the operation behavior feature sequence group.

[0033] The path direction change recognition submodule calls the operation behavior feature sequence group to analyze the correspondence between the direction number and the click position change during the page jump process, identify the position where the forward and backward directions change in the path segment, extract the click action where the direction is inconsistent with the target position, and obtain the direction alternation behavior index set; First, obtain the following three items from each action sequence: page jump direction number, click position number, and page number. The direction number can be set to forward or reverse based on the order of the starting and ending page numbers. If the previous page number is P1 and the next page number is P2, and P2 is greater than P1, it is marked as forward; otherwise, it is marked as reverse. The click position number is the target element number recorded in the click event, and the page number is the source or target page identifier of the jump action. After extracting the above three items, establish a one-to-one correspondence between the direction number and the click position number, and process them sequentially according to time. The behavior sequence is analyzed to determine if there is a change in direction number between two adjacent jump actions. For example, if two consecutive jump operations in the behavior sequence change direction from forward to reverse, the point is considered a direction switch point. At this time, it is necessary to simultaneously determine whether the click position number at this location has changed. If the click position number is different from the previous one, it is considered an event where the jump direction and the click target are inconsistent. Conversely, if the click position number remains the same, but the direction changes, it is necessary to check whether different interactive modules are loaded on the page content at this node. If the interactive module identification codes are different, it is still marked as a target change action. The number of such events where the jump direction and target position are not synchronized in the entire behavior sequence is further checked. An index number is established for each location where a direction change occurs, and the click position where the target change first occurs in the subsequent jump action is used as the positioning basis to form a corresponding structure of direction jump and target position change. Then, these indices are output sequentially according to the order of the behavior occurrence time. During the process, it is necessary to exclude jump behavior interference caused by automatic page refresh or pop-up prompts. This can be eliminated by using the "Source Type" of the event attribute field as "System Triggered" to ensure that only the direction change and target position caused by user active operation are retained. Misaligned behavior, for example, in a sequence of actions, a click action numbered N1 triggers a jump from page A to page B in a forward direction, and then a click action numbered N2 triggers a jump from page B to page A in a reverse direction. If the click positions of N1 and N2 are numbered L1 and L3 respectively, and are not the page back button, then it can be determined that the direction is inconsistent with the target. In this sequence, this position is the identified valid alternating behavior point. Following this logic, all action sequences are traversed, all operation points that meet the above conditions are extracted and their action numbers are recorded, and summarized into an index set to obtain the alternating direction behavior index set.

[0034] The offset segment corresponding mapping submodule extracts the click nodes between the task abandonment position and the jump start position based on the direction alternation behavior index set, compares the operation corresponding to the node in the segment with the original action sequence in the path, and obtains the path structure reconstruction fragment set. First, it is necessary to define the behavioral path segment between the jump start position and the task abandonment position corresponding to each record. Extract all user-triggered click node numbers from this path segment. These node numbers come from the click target field in the event log, representing the interactive area operated by the user on a specific page, such as buttons, tabs, navigation points, etc. During the extraction process, click behaviors automatically triggered by the system need to be excluded. This can be identified and eliminated by the field in the click event attribute where the "trigger source" is "system automatic". Only the node records actively triggered by the user are retained. Then, the node numbers are sorted according to the trigger time field of each node event to form a node sequence corresponding to the time sequence. In this sequence, the page number, node number and trigger order of each node are recorded. Then, each node operation in the path segment is compared with the preset operation action order in the original path. The original action order can be read from the path structure initialization template. This template records the recommended click order number under each page. For example, the recommended operation order in page X is N1-N2-N3. If the order N1-N3-N2 appears in the user behavior path, it can be determined that the user has deviated from the node jump order in the path. In an educational context, this offset often signifies a mismatch between the student's default learning path and their actual knowledge exploration needs based on their cognitive abilities. By combining this with the student's learning objective data, it can be determined whether they are lost or actively expanding their learning. Further, the sequence offset points between the user's behavioral path and the original path are extracted. Action mapping is performed in areas where node order is misaligned. For example, if a user clicks N4-N5-N6 in page B, while the original path should be N4-N6-N5, this segment can be extracted as an operation offset fragment. Combining the node mapping results between the task abandonment position and the jump start position, the path segment corresponding to the misaligned action segment is numbered and aligned with the original path segment. For all path segments meeting the above conditions, their click node sequence, sequence misalignment position, and corresponding page identifier are recorded sequentially. Finally, based on the relative position of the node misalignment and the node calls before and after the jump, the actual execution fragments of the user's offset path segment in terms of operation order are compiled and numbered chronologically into a set, resulting in a path structure reconstruction fragment set.

[0035] Specifically, such as Figure 2 , 6 As shown, the teaching node jump path construction module includes: The path segment extraction submodule reconstructs the segment set based on the path structure, extracts the node number and content tag corresponding to the operation behavior in the path, compares the number of repetitions of the node number, removes the number of repetitions exceeding the preset frequency threshold, and obtains the path behavior time series node number. First, the node number and associated content tag corresponding to the operation behavior are read from each segment. The node number can be extracted from the target identifier field of the click event in the behavior record, and the content tag comes from the teaching content attribute field bound to the node in the original task configuration. For example, a node number is N7, and its corresponding tags are "Knowledge Point A" and "Practice Module B". After reading, a one-to-one mapping structure between node numbers and tags is established. Then, the frequency of each node number in the entire segment set is summarized, and the frequency of all numbers is entered into a statistics list. In this statistical process, the node number and its segment number, time order, and adjacent behavior numbers in each path segment need to be recorded. If a number appears more than three times in multiple segments and its behavior characteristics are consistent, such as the time interval between clicking and jumping to the target page being the same as the action trigger time, it is recorded as a repeated behavior node. Then, the number of repetitions is judged by node number. If the number of times a number is recorded in non-contiguous path segments accounts for more than 20% of the total number of segments, it can be regarded as a concentrated occurrence number. This judgment criterion comprehensively considers the distribution patterns of the total number of path segments and node numbers to eliminate local repetitions caused by operational inertia. Subsequently, the identified numbers are removed from the original number set by skipping the corresponding behavior record and not adding its behavior field to subsequent paths. The remaining number sequence is arranged according to its original chronological order in the behavior log, constructing a time-series number sequence structure. The operation behavior and content tag of each node in the sequence are jointly written into the output structure, while the path segment number, click order, and tag field value are recorded at each node position. After the entire processing is completed, the path behavior time sequence node numbers are obtained.

[0036] The tag matching and recognition submodule calls the node tags in the path behavior time series node numbers, matches each item in the content tag set with the behavior tag, identifies the set of node tags that do not appear in the path, filters the tag items with the same number of repetitions, and locates the corresponding node number to obtain the set of node numbers with the same number of tag repetitions; First, the tag set corresponding to each node number is read. This set can be extracted from the original node configuration. Each tag is a teaching concept identifier bound to the content domain. For example, node N1 is bound to tags T1 and T3, and node N2 is bound to tag T2. Then, a complete set of behavior tags is constructed. Each item in the set is stored with the node number as the key and the tag set as the value. On this basis, a complete set of content tags is introduced as a matching reference set. The content tag set comes from all the tag items preset in the course resource structure definition. Its format is a set structure from T1 to Tn. In the matching operation, each item in the behavior tag set is traversed, and the number of times each tag in the behavior tag appears in the content tag set is recorded. At the same time, the number of times each tag item appears in the behavior tag set is counted in the content tag set. The corresponding results are used to build a tag frequency lookup table according to the tag item number. For example, T1 appears 3 times in the behavior tag set, T2 appears 2 times, and so on. On this basis, tag items that do not appear in the behavior path are further identified, that is, the behavior tag set is filtered out. Uncovered content tags form a subset of unmatched tags. For example, if the content tag set is T1 to T6, but the behavior tag set only contains T1 to T4, then T5 and T6 are the missing items. Then, for each missing tag item, the node number bound to it in the content resource is searched to establish a correspondence between tags and node numbers, and its identifier position in the resource structure is recorded. Next, the number of tags bound to the corresponding node in the resource structure for each tag item is compared, and tag items whose number of occurrences matches the number of occurrences in the behavior tag set are extracted. Their corresponding node numbers are then filtered out. For example, if T2 appears twice in the behavior path, and T5 also has two tags bound to it in the resource structure with the same tag number as T2, then the node number corresponding to T5 is included in the result range. Finally, a node number filtering structure is constructed using the tag number, the number of occurrences, and the corresponding node number as the three primary keys. Numbers that have already appeared in the behavior path are removed, and only the set of node numbers that have not appeared in the behavior path is retained and written into the output field, resulting in the set of node numbers with the same tag repetition count.

[0037] The jump node sequence output submodule is based on the node number set of nodes with the same number of repetitions of tags. It compares the number range of tags corresponding to the content tags, selects the content item numbers with the same number of tags, and extends the node number sequence according to the time order of the path behavior time series node numbers to obtain the teaching jump node sequence group. First, the number of tags bound to each ID in the set is extracted. This information can be obtained from the resource configuration table. The tag field of each content item lists the total number of tags associated with it. For example, if ID N8 is bound to two tags, T1 and T2, then its tag count is 2. Next, learning progress data is introduced. In this progress table, each content item contains a tag count field, which indicates the tag capacity to be processed in the current learning stage. The tag counts of each content item in the learning progress are traversed, and IDs with the same number of tags bound to the above node IDs are selected. For example, if ID M5 in the learning progress corresponds to 2 tags, which is the same as N8, then ID M5 is retained, forming a set of content item IDs with the same number of tags. Next, this set is matched one-to-one with the IDs in the set of node IDs with the same number of tag repetitions, forming optional jump node ID pairs. Then, the node order recorded in the path behavior time series node IDs is called. This order is based on the behavior occurrence time. The nodes are arranged, such as N1 to N7, corresponding to different operation timestamps. A time series vector is constructed accordingly. The numbering of content items with the same number of tags is matched and inserted into the end of the time vector to extend the numbering structure of the time series. During the insertion process, it is necessary to check whether there are duplicate numbers. If they already exist in the time series, they are skipped to ensure the uniqueness of the extended numbers. At the same time, the number of tags corresponding to the newly added number and the insertion position are recorded. If two newly added numbers have the same number of tags and are inserted in consecutive positions, they can be grouped into the same jump path segment. Finally, the extended numbering structure and the set of newly added node identifiers are combined into the output structure. Based on the original time series and supplemented by extended nodes, a complete behavioral path trajectory is formed. This is used for subsequent cross-validation with objective factors such as students' learning objectives, learning status, and cognitive abilities to realize the dynamic allocation of teaching resources and the construction of jump logic that meets their real cognitive needs, resulting in a teaching jump node sequence group.

[0038] Specifically, such as Figure 2 , 7 As shown, the teaching task configuration generation module includes: The task parameter extraction submodule is based on the teaching jump node sequence group. It extracts the task style, resource presentation method and interaction flow associated with the node, and extracts the resource structure corresponding to the task style according to the node order, corresponding resource number and task style identifier, to obtain the node task parameter merge list. First, read the node number in the sequence and then sequentially access its corresponding task style field, resource presentation method field, and interaction flow field in the task configuration table. The task style field describes the question type or learning activity type corresponding to the node, such as multiple choice, fill-in-the-blank, or practice exercises. The resource presentation method field describes the display format of the materials on the page, such as text, images, or multimedia combinations. The interaction flow field describes the order of the user's actions within the node, such as reading before clicking or selecting before confirming. During the reading process, the node number order must be kept unchanged to ensure that the subsequent parameter correspondences are clear. Then, for each node, the task style is matched with its associated resource configuration item. The resource configuration item includes a resource number field to uniquely identify the specific content material. For example, resource number R3 represents a section of explanatory text, and resource number R7 represents a practice page. During processing, the node number, task style identifier, and resource number are recorded in a three-way correspondence to form a node-level parameter entry. For example, node N2 corresponds to task style S1, resource number R3, and the interaction flow is F2. This entry serves as a basic parameter unit. The above reading and matching process is then repeated for the next node in the sequence, gradually forming... This process involves creating a parameter set covering the entire sequence of jump nodes. It's crucial to distinguish between different resource IDs associated with the same task style across different nodes. For example, two nodes might both be multiple-choice questions, but with corresponding resource IDs R4 and R9 respectively. These should be recorded as separate entries. Then, all acquired node parameter entries are sequentially integrated, arranging them solely based on their position in the jump sequence without introducing additional sorting rules. This ensures the parameter order matches the user's actual jump path. Within this order, the task style identifier for each node is used as the primary key, and the resource ID and interaction flow identifier are used as secondary keys. Fields are written to the same record row, for example, forming record N5-S2-R8-F1. Such records are stored consecutively according to the node order, forming a complete parameter list. In a practical application example, if the teaching jump node sequence is N1, N3, N5, then the explanation style and text resources corresponding to N1 are read in sequence, the practice style and question resources corresponding to N3 are read, and the feedback style and prompt resources corresponding to N5 are read. Finally, a set of node parameters arranged in the path order is formed. After the parameters of all nodes are extracted and mapped, the set is written to the output area as the input basis for subsequent task execution and resource scheduling, resulting in a merged list of node task parameters.

[0039] The operation style mapping submodule extracts the sequence of executed operation methods based on the task styles and interaction flow content in the node task parameter merge list, compares the question structure with the material format style, and marks the related operation styles to the corresponding nodes to obtain the node operation style mapping structure. First, extract the task style identifier and corresponding interaction step number registered in the list for each node. The task style indicates the type of teaching task presented by the node, such as true / false questions, drag-and-drop sorting, and image-text fill-in-the-blank. The interaction flow indicates the operation process that the user needs to go through during the task execution, such as multiple specific action types such as clicking, swiping, dragging, and inputting. Then, read the sequence of operation methods recorded by the user in the same time window. This sequence comes from the interaction events collected in the behavior log and includes fields such as click position number, operation target identifier, and event triggering order. For example, if a user performs click on the button, drag the block, and click confirm in sequence on page N3, it is recorded as three operation methods: O1, O3, and O5. Next, compare the operation methods performed by the user in chronological order with the interaction flow in the task parameters one by one. For example, if the task style S2 is "drag-and-drop sorting", its standard interaction flow includes operation methods O3 and O4. If O3 exists in the user's sequence, it is considered to be associated with the task style. Continue to compare the matching relationship between the question type structure and the material format. The question type structure is provided by the structure definition field of the task style. For example, the structure of S2 is defined as "area positioning + sequential operation". The material form is provided by the content material type described in the resource presentation field, such as blocks, text boxes, and blank placeholders. If the click target number involved in the user operation falls exactly in the block area and the sequence matches the structure definition, the matching status of the operation method and the task style of the node can be further confirmed. The operation method that meets the matching conditions will be marked as the valid style of the current node, and the operation method number will be written into the operation style field of the node. For example, node N3 is marked with O3 as its task style matching operation. If the same operation method is detected in multiple nodes, it is necessary to determine whether the method meets the task style requirements to avoid mislabeling. For example, O1 may be a confirmation operation. If it does not appear in the task style interaction flow, it will not be marked. Finally, all matched nodes are summarized according to the node number. For each node, three items are written: task style identifier, matching operation method, and interaction flow program number. The association relationship is established according to the node order to obtain the node operation style mapping structure.

[0040] The execution method analysis submodule compares the content direction and task style differences between adjacent nodes based on the node operation style mapping structure, groups them into stages according to the jump path, detects the distribution of task styles in each group, and obtains the set of teaching task execution methods. First, extract the sequential number, task style name, and marked operation style of each node in the behavior path. Then, based on the connection relationship between nodes in the jump path, identify the page jump direction between adjacent nodes in sequence. The page direction is determined according to the page number order. If the page number of the later node is greater than that of the earlier node, it is marked as a forward jump; otherwise, it is a reverse jump. Next, extract the task style name for each pair of adjacent nodes and compare it to determine if the task style has changed. For example, if node A corresponds to a true / false question and node B corresponds to a fill-in-the-blank question, then the task type of this jump segment has changed. Record the node numbers of such changes and organize them into a difference identifier list. Then, divide and group the complete path according to the continuity of the jump direction and whether the task type is consistent. When adjacent nodes have the same direction and the task type remains unchanged, they are grouped into the same stage group. If the jump direction or the task type changes, a new stage group is redefined. For example, if the jump order from node 1 to node 2 is consistent in direction and the task type is multiple choice, then it forms the first stage group. If the direction of node 3 is reversed or the task becomes a sorting task, then it is divided into the second stage group. Each group includes node number, task type, and operation style. Based on this, the frequency of each task type in each group is counted. For example, if a stage group contains five nodes, three of which are true / false questions and two are drag-and-drop tasks, then the task type distribution of this group is: true / false questions appear three times and drag-and-drop tasks appear twice. If a task type does not appear repeatedly in a group, such as each task type appearing only once, it is marked as a mixed task stage. The dominance of task types in consecutive groups is further analyzed. If a certain type appears frequently in more than three consecutive stage groups, it is recorded as a continuously dominant type. The operation style number in each stage is further associated to identify whether the operation method changes with the task type. If a certain type always corresponds to the same operation method, such as click selection, a stable operation style mark is formed. If the task type is the same but the operation style changes, it is recorded as an operation style differentiation node. The entire path stage structure is organized to form a four-item linked structure set of stage number, task type, operation style, and direction order. This is written into the structure table in the stage order. Finally, combined with the aforementioned input objective cognitive data, accurate push is completed to obtain the set of teaching task execution methods.

[0041] 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 variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should 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 protection of the described technical solutions.

Claims

1. An adaptive teaching system based on learning behavior analysis, characterized in that, The system includes: The learning focus module obtains the page dwell order, click jump order and operation trajectory, tracks page switching and position changes, divides the operation sequence into segments, and obtains a list of operation focus behavior segments; The operation segment interference identification module, based on the operation focused behavior segment list, compares the question type structure with the operation sequence, tracks the changes in actions before and after the jump behavior, and obtains a set of behavior interference segment identifiers. The behavior offset segment identification module extracts jump behavior, misdirected click and back point based on the behavior interference segment identifier set, compares the direction change and action repetition, and obtains the path structure reconstruction segment set according to the order between the task abandonment position and the jump start position. The teaching node jump path construction module reconstructs the fragment set based on the path structure, extracts duplicate label nodes, finds label nodes that do not appear in the path, and advances the jump path according to the time sequence of the operation behavior to obtain the teaching jump node sequence group. The teaching task configuration generation module extracts task styles, resource methods and interaction flows based on the teaching jump node sequence group, compares the task content of adjacent nodes, analyzes the distribution of tasks in the path, and obtains a set of teaching task execution methods.

2. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The list of operation-focused behavior segments includes page dwell order, click jump order, operation path trajectory, page switching frequency, click position repetition, and time sequence arrangement. The set of behavior interference segment identifiers includes jump behavior position distribution, question type structure differences, operation sequence interruption points, page jump direction changes, and operation chain interruption nodes. The set of path structure reconstruction fragments includes jump behavior sequences, misdirected click actions, page back points, click direction change order, click target inconsistency segments, and the mapping relationship between task abandonment position and jump start position. The teaching jump node sequence group includes repeated label nodes, non-appearing node set, label-covered content items, label name correspondence, and time sequence extension nodes. The set of teaching task execution methods includes task style type, resource presentation form, interaction flow structure, question type structure mapping method, content direction change characteristics, and task style distribution.

3. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The question structure refers to the question types and structural features bound to each learning page or node; The task abandonment location refers to the node and page where a student leaves without completing a learning task due to interruption, exit, or incorrect clicks.

4. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The jump start position refers to the starting point from which the student begins the jump operation on the page; The resource method refers to the presentation format of the teaching content on the page.

5. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The learning-focused behavior characterization module includes: The page behavior sequence acquisition submodule acquires the order of student's page stay, click jump order and operation path trajectory in the question reading area, option browsing area and submission operation area. Based on the time sequence of page stay actions and click actions, it extracts the action sequence of page switching and position change, analyzes the order of actions on the timeline, and obtains the page operation behavior sequence. The operation trajectory segmentation submodule, based on the page operation behavior sequence, extracts the behavior trajectory where the page does not switch but the click position is continuously repeated according to the correspondence between the page dwell position and the time of the click action, and draws out the operation segments where the page does not change continuously, thus obtaining the operation trajectory segment set; The focus segment extraction submodule extracts segments with continuous page switching based on the operation trajectory segment set, and obtains a list of operation focus behavior segments by corresponding to the page number and click position number in each segment.

6. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The operation segment interference identification module includes: The path behavior structure comparison submodule extracts the question type structure number, page dwell order and operation click order number in the previous and subsequent paths based on the operation focus behavior segment list, and compares the positions in the path where the order of appearance of question type structure and operation order change to obtain the question type behavior interruption interval set. The page jump direction analysis submodule extracts the starting page and target page direction numbers based on the question type behavior interruption interval set, compares the page jump direction changes in the path segment, identifies operation segments where the jump direction is repeatedly switched between adjacent positions, and obtains the page jump direction change segment set. The interference node extraction submodule extracts the click node numbers between the first time the page jumps out and the last time the page is returned during the jump operation based on the set of page jump direction change segments. The click nodes within the segments are then aggregated to obtain a set of behavior interference segment identifiers.

7. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The behavior offset segment identification module includes: The behavior action sequence extraction submodule extracts jump behaviors, misdirected click actions, and page back points within the path segment based on the behavior interference segment identifier set. It then identifies repeated behaviors by corresponding page numbers, click positions, and direction changes in time sequence to obtain a set of operation behavior feature sequences. The path direction change recognition submodule calls the operation behavior feature sequence group to analyze the correspondence between the direction number and the click position change during the page jump process, identifies the position where the forward and backward directions change in the path segment, extracts the click action where the direction is inconsistent with the target position, and obtains the direction alternation behavior index set; The offset segment corresponding mapping submodule extracts the clicked nodes between the task abandonment position and the jump start position based on the direction alternation behavior index set, compares the operation corresponding to the node in the segment with the original action sequence in the path, and obtains the path structure reconstruction fragment set.

8. The adaptive teaching system based on learning behavior analysis according to claim 7, characterized in that, In the process of extracting jump behavior, misdirected click action and page back point within the path segment: within the page range covered by the same behavior interference segment identifier set, identify click actions with consecutive time intervals, classify the jump behavior to which the consecutive clicks belong, and extract click actions with reverse direction offset according to the spatial change trend between the click position and the page jump direction. In the process of identifying recurring behaviors: in the group of operation behavior feature sequences, the correspondence between page number and click position is compared, and the same operation behavior features that recur in the position sequence are filtered. During the process of extracting the click nodes between the task abandonment position and the jump start position: based on the jump start point shown in the direction alternation behavior index set, trace forward along the user operation sequence and extract the click nodes in the path after the jump start point.

9. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The teaching node jump path construction module includes: The path segment extraction submodule reconstructs a segment set based on the path structure, extracts the node numbers and content tags corresponding to the operation behaviors in the path, compares the number of times the node numbers are repeated, removes the numbers whose number of repetitions exceeds a preset frequency threshold, and obtains the path behavior time series node numbers. The tag matching and recognition submodule calls the node tags in the path behavior time series node numbers, matches each item in the content tag set with the behavior tag, identifies the node tag set that does not appear in the path, filters the tag items with the same number of repetitions, and locates the corresponding node number to obtain the node number set with the same number of tag repetitions; The jump node sequence output submodule compares the node numbers in the set of node numbers with the same number of repetitions of the tags with the range of the number of tags corresponding to the content tags, selects the content item numbers with the same number of tags, and extends the node number sequence according to the time order of the path behavior time sequence node numbers to obtain the teaching jump node sequence group.

10. The adaptive teaching system based on learning behavior analysis according to claim 1, characterized in that, The teaching task configuration generation module includes: The task parameter extraction submodule extracts the task style, resource presentation method and interaction flow associated with the node based on the teaching jump node sequence group. It extracts the resource structure corresponding to the task style according to the node order, corresponding resource number and task style identifier, and obtains the node task parameter merge list. The operation style mapping submodule extracts the sequence of executed operation methods based on the task styles and interaction flow content in the node task parameter merge list, compares the question structure and material format style, and marks the related operation styles to the corresponding nodes to obtain the node operation style mapping structure. The execution method analysis submodule compares the content direction and task style differences between adjacent nodes based on the node operation style mapping structure, groups them into stages according to the jump path, detects the distribution of task styles in each group, and obtains the set of teaching task execution methods.

Images