Collaborative dictation and group error pattern analysis method and system

Abstract

This application provides a collaborative dictation and group error pattern analysis method, including: performing temporary error identification, cluster similarity classification, and association strength identification on the data in the group dictation feature dataset to obtain a group error map; quantifying and visualizing the common error features and error association features in the group error map to generate a teaching insight dashboard with standardized group learning data; using the standardized group learning data from the teaching insight dashboard as input, processing it through a preset teaching material generation algorithm to obtain a group review material package and target learning quantitative indicators matching the group review material package; applying the group review material package to group users in a class or group, obtaining secondary learning behavior data after user use, and analyzing it to obtain the resulting learning quantitative indicators for iterative optimization.

Description

Technical Field

[0001] This application relates to the field of educational technology, and more specifically, to a collaborative dictation and group error pattern analysis method and system. Background Technology

[0002] Dictation, as a core component of language teaching to solidify foundations and assess student learning, presents numerous challenges in traditional classroom teaching. Currently, teachers largely rely on manual grading of dictation assignments, which is not only time-consuming and labor-intensive but also fails to quickly collect and analyze the error characteristics of the entire class. Identifying common weaknesses and frequently made mistakes depends entirely on subjective experience, lacking objective data support, and is highly likely to overlook persistent knowledge gaps.

[0003] In response to the identified errors, the review materials compiled by teachers are mostly homogeneous exercises and repetitive dictation content, lacking targeted adaptation to the learning situation of the group. More importantly, the existing teaching model lacks a quantitative and logical judgment mechanism: it is impossible to count the error correction rate of students after review, it is difficult to distinguish between temporary mistakes and persistent difficulties, and it is impossible to objectively determine the actual effectiveness of the review plan. Once the review materials are generated, they remain fixed and cannot be dynamically optimized according to the real-time learning situation.

[0004] If this continues, common errors among students will not be addressed in a targeted manner, individual differences in learning will be ignored, teaching decisions will become disconnected from actual learning situations, review efficiency will remain low, precise teaching will be difficult to implement, and the needs of large-scale, data-driven modern teaching will not be met. Summary of the Invention

[0005] The main purpose of this application is to provide a collaborative dictation and group error pattern analysis method and system to solve the technical problems in the background art.

[0006] To achieve the above objectives, the first aspect of this application proposes a collaborative dictation and group error pattern analysis method and system, comprising: Acquire learning behavior data of a group of users within a class or group, and extract features to obtain a group dictation feature dataset; The data in the group dictation feature dataset are subjected to temporary error identification, cluster similarity classification, and association strength identification operations in sequence to obtain a group error map, wherein the group error map represents the common error features and error association features of the group; The common error features and error correlation features of the group in the group error map are quantified and visualized to generate a teaching insight dashboard with standardized group learning data; Using the standardized group learning data from the teaching insight dashboard as input, the data is processed through a preset teaching material generation algorithm to obtain a group review material package and target learning quantitative indicators that match the group review material package. The group review material package is applied to the group users in the class or group, and the secondary learning behavior data of the users after use is obtained. The secondary learning behavior data is analyzed to obtain the result of quantitative indicators of learning progress. Based on the comparison between the quantitative indicators of the learning situation and the quantitative indicators of the target learning situation, the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for the identification of association strength, and the logic of the preset teaching material generation algorithm are iteratively optimized to achieve targeted generation of dictation error patterns and review materials for group users within a class or group.

[0007] Secondly, this application provides a collaborative dictation and group error pattern analysis system, applied to the aforementioned collaborative dictation and group error pattern analysis method, comprising: The acquisition unit is used to acquire learning behavior data of a group of users within a class or group, and to extract features to obtain a group dictation feature dataset. The processing unit is used to perform temporary error identification, cluster similarity classification, and association strength identification on the data in the group dictation feature dataset in sequence to obtain a group error map, wherein the group error map represents the common error features and error association features of the group; The quantization unit is used to quantify and visualize the common error features and error correlation features in the group error map, and generate a teaching insight dashboard with standardized group learning data. The generation unit is used to take the standardized group learning data of the teaching insight dashboard as input, process it through a preset teaching material generation algorithm, and obtain a group review material package and a target learning quantitative indicator that matches the group review material package. The verification unit is used to apply the group review material package to the group users in the class or group, obtain the secondary learning behavior data of the users after use, and analyze the secondary learning behavior data to obtain the result learning quantitative indicators. The results unit is used to iteratively optimize the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for association strength identification, and the logic of the preset teaching material generation algorithm based on the comparison difference between the result learning quantitative indicators and the target learning quantitative indicators, so as to achieve targeted generation of dictation error pattern analysis and review materials for group users within a class or group.

[0008] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the aforementioned collaborative dictation and group error pattern analysis method.

[0009] Fourthly, this application provides a computer program that, when executed by a processor, implements the steps of the aforementioned collaborative dictation and group error pattern analysis method.

[0010] The technical solutions provided by the embodiments of this application may include the following beneficial effects: This application presents a collaborative dictation and group error pattern analysis method and system that effectively addresses the technical pain points of traditional dictation teaching, such as time-consuming and labor-intensive manual grading, lack of objective data support for group error analysis, homogenized and untargeted review materials, lack of quantitative judgment mechanisms, and inability to dynamically optimize teaching strategies. It possesses significant technical effects and teaching application value: First, by automating the acquisition of group learning behavior data and performing standardized error judgment operations, it replaces manual collection, sorting, and analysis of group error characteristics, objectively generating error maps representing common group error characteristics and correlations. This transforms learning situation assessment from subjective experience-based judgment to data-driven analysis, significantly improving the efficiency of learning situation assessment and freeing teachers from tedious manual statistical work. Second, based on quantitative data of group learning situation, it automatically generates targeted group review material packages, matched with corresponding target quantitative indicators of learning situation, fundamentally avoiding the problem of homogenized review materials. This achieves precise adaptation of review materials to the learning situation of the class / group, improving the effectiveness of review content. The method offers several key advantages: First, it is highly targeted. Second, by acquiring secondary learning behavior data, it obtains quantitative indicators of learning progress and establishes a quantitative comparison and judgment mechanism between the results and target learning progress indicators. This mechanism can accurately count the error correction rate after review and effectively distinguish between temporary mistakes and persistent difficulties based on the preliminary temporary error identification operation, thus objectively judging the effectiveness of the review plan. Third, based on the comparative differences of learning progress indicators, it iteratively optimizes the rules, thresholds, and teaching material generation algorithm logic for error judgment. This allows the logic of dictation error analysis and review material generation to dynamically adjust with the real-time learning progress of the group, ensuring that teaching decisions are highly matched with actual learning progress and overcoming the problems of fixed review materials and disconnect between teaching and learning progress in the traditional model. Fourth, this method adopts a standardized and data-driven processing approach throughout the entire process, which can adapt to the group analysis needs of multiple classes and groups. It has the technical foundation for large-scale application, effectively promotes the implementation of precision teaching, improves the overall efficiency and review effect of dictation teaching, and meets the application needs of modern teaching on a large scale and with data. Attached Figure Description

[0011] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 A flowchart of a collaborative dictation and group error pattern analysis method provided in this application. Detailed Implementation

[0012] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0013] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0014] like Figure 1 As shown, in a first aspect, this application provides a collaborative dictation and group error pattern analysis method, including: S100: Obtain learning behavior data of a group of users within a class or group, and extract features to obtain a group dictation feature dataset.

[0015] The learning behavior data of a class or group of users is acquired and features are extracted to obtain a group dictation feature dataset. The implementation logic is as follows: first, the entire dictation process data of the group users is collected; then, data interference is eliminated through preprocessing, and a unified format is applied; finally, core features of individual and group dimensions are extracted hierarchically and integrated in a structured manner to form a standardized dataset that can be directly used for group error pattern analysis.

[0016] Specifically, obtaining a group dictation feature dataset may include the following steps: S101: Obtain dictation-related learning behavior data from group users within a class or group.

[0017] Specifically, the system scans dictation papers, homework assignments, and other written materials used by users during their learning process, taking classes or groups as units. It then performs data conversion on the scanned documents to collect and label dictation-related learning behavior data, covering the entire process of dictation tasks from "answering to grading to reviewing".

[0018] Data collection and conversion may include scanning dictation papers or assignments to form PDF or image data, and then converting them into analyzable word, sentence, or passage spelling or handwritten text data using conventional OCR and other technologies; Basic error annotation: For the converted text data, basic error annotation is performed using conventional natural language processing methods to form error annotation data, which marks the error location and error type (spelling and grammar). Privacy and data association guidelines may include assigning a unique anonymous identifier to each user, replacing privacy information such as name and student ID, while retaining "class or group affiliation + anonymous ID"; assigning a unique task ID to each dictation task, and storing the written text data and basic correction and annotation data together with this ID for easy analysis by task dimension later. The final output consists of raw data for dictation-based learning behaviors, including written text data and basic correction and annotation data.

[0019] S102, the dictation learning behavior data is preprocessed by sequentially performing data cleaning, outlier removal and normalization to obtain preprocessed data.

[0020] Specifically, the data collected in step S101 undergoes preprocessing including data cleaning, outlier removal, and normalization. It should be noted that data cleaning, outlier removal, and normalization are standard data processing operations, and this application does not represent any improvement to the technical solution for data processing.

[0021] S103, extract and structure-integrate the dictation error features, answer behavior features and common features of group answers from the preprocessed data to obtain a group dictation feature dataset.

[0022] Specifically, the basic features of dictation errors and answering behavior at the individual user level are first extracted, then statistically aggregated based on individual features into common features of group answering, and finally all features are integrated according to a unified structure and format to form a standardized group dictation feature dataset that can be directly connected to subsequent group error pattern analysis.

[0023] Feature extraction is as follows: Dictation error characteristics (individual dimension): Extracting the user's objective error characteristics: error type (spelling, grammar), number of errors, error location (first / last letter of a word, predicate of a sentence), and accuracy of knowledge points; Answering behavior characteristics (individual dimension): Extract normalized objective behavioral numerical characteristics of users: number of modifications, and the fluctuation range of accuracy in multiple dictation tests (reflecting answering stability). Common characteristics of group responses (group dimension): Based on statistical aggregation of individual characteristics, extract high-frequency error types of the group (such as 50% of users in the class making "spelling error at the end of -ed") and co-occurrence combinations of different error types (such as tense error + preposition error).

[0024] The structured integration is as follows: The unified dataset has a three-layer structure: 1. Basic information layer (task ID, class identifier, number of users, collection time); 2. Individual feature layer (identified by anonymous user ID, integrating individual error and behavioral features); 3. Group feature layer (identified by class identifier, integrating common group features, and associating with the basis for individual feature aggregation). Unified storage format: Numerical features can retain two decimal places, categorical features use standard numeric encoding (e.g., "spelling error" is encoded as 01), and text features use structured fragments (e.g., "error location: word ending in -ed"). The final result is a group dictation feature dataset that can be directly used for group error pattern analysis.

[0025] S200, the data in the group dictation feature dataset are subjected to temporary error identification, cluster similarity classification and association strength identification in sequence to obtain the group error map.

[0026] The group error map represents the common error characteristics and error correlation characteristics of the group.

[0027] Using the group dictation feature dataset output by S103 as input, the system first performs temporary error identification through preset rules, eliminating false error data with no analytical value, and obtaining a group true error dataset that only reflects the mastery of knowledge points. Then, a clustering similarity judgment threshold is set, and the group true error dataset is calculated and clustered according to error type and error cause, and the frequency of occurrence of each type of error in the group is counted. Finally, the correlation strength is used to identify and mine the correlation between errors, and integrate them to form a group error map that represents the common error characteristics and error correlation characteristics of the group, providing core analytical basis for subsequent teaching decisions and the generation of review materials.

[0028] Specifically, obtaining a population error map may include the following steps: S201, based on the preset rules for identifying typos, the error data in the group dictation feature dataset is identified, and the false error data corresponding to temporary typos is removed to obtain the group true error dataset.

[0029] Specifically, based on the structured error types, error locations, and knowledge point accuracy rates in the group dictation feature dataset, temporary pen error identification rules are preset. By matching these rules, false errors (temporary pen errors or operational mistakes that do not reflect insufficient knowledge mastery) and true errors (caused by a lack of knowledge mastery, which are the core analysis objects) are distinguished. After batch removal of false error data, the remaining true error data is structured and integrated to obtain a clean group true error dataset, which eliminates invalid data interference for subsequent clustering and classification.

[0030] For example, a pre-defined unified temporary error identification rule is constructed for the group dictation feature dataset output by S103. The rule design fits the error / operational mistake features of the dictation scenario. All rules are based on the feature settings in the group dictation feature dataset without subjective assumptions. If any rule is met, it is judged as a temporary error (false error). If none of the rules are met, it is judged as a true error.

[0031] The rules are as follows: Rule 1: Edit Distance Judgment Rule: For word-based dictation errors, calculate the edit distance between the user's incorrectly spelled text and the standard text. If the edit distance is ≤1, it is judged as a temporary typo. Rule 2: Format difference judgment rule: If the error is only manifested as a difference in capitalization or the addition or deletion of spaces, without any substantial spelling or grammatical error, it is judged as a temporary typo; Rule 3: Recognition Error Judgment Rule: If the error is caused by OCR scanning recognition error, and the original writing medium can be clearly identified as the correct answer, it is judged as a temporary pen error; Rule 4: Low-frequency isolated error judgment rule: If a user's error on a certain knowledge point occurs only once, and the overall accuracy rate of that knowledge point is ≥95%, it is judged as a temporary typo.

[0032] From the group dictation feature dataset, extract feature subsets related to error data, group them according to the dimensions of "user anonymous ID - task ID (number of each dictation task) - knowledge point - error type", and retain the core features of each group, such as error location, knowledge point accuracy rate, and comparison results between the error text and the standard text. Then, match each of the above four discrimination rules sequentially, labeling each rule with "rule matching result" (match / not match). Based on the rule matching results, if any rule is met, it is judged as a false error (temporary typo); if none are met, it is judged as a true error, and the final error nature is labeled for each piece of error data. All data labeled as "false errors" are uniformly filtered and batch-removed. Finally, retain all data labeled as "true errors", restructure them according to the dimensions of "class or group identifier - task ID - knowledge point - error type", retain all related identifiers and core features of the original data, ensuring that the data structure is consistent with the group dictation feature dataset, and only remove false error data to form the group true error dataset.

[0033] S202, set a clustering similarity judgment threshold, perform similarity calculation and clustering classification on the group true error dataset according to error type and error cause, and count the frequency of group occurrence of each type of error to obtain group error classification feature data.

[0034] Specifically, based on the structured features of the group true error dataset, a reasonable clustering similarity judgment threshold is first set as the core basis for error classification. Then, from the two core dimensions of error type and error cause, the similarity of all true error data is calculated, and errors with similarity ≥ the judgment threshold are classified into the same category. Finally, the frequency of occurrence of each type of error is statistically analyzed from both individual and group dimensions, and the clustering results are structurally integrated to obtain group error classification feature data, laying the foundation for subsequent error association strength identification and group error map construction.

[0035] For example, the threshold setting is based on the error feature distribution pattern in the dictation scenario, taking into account both the accuracy and rationality of classification. The cluster similarity judgment threshold is set at 80%, and the threshold can be flexibly adjusted according to the dictation task type (word dictation / sentence dictation), with an adjustment range of 75%-85%. If the similarity is below 75%, errors with different causes are easily classified into the same category, reducing classification accuracy; if the similarity is above 85%, similar errors are easily over-segmented, failing to reflect the common error characteristics of the group. 80% is the optimal judgment threshold.

[0036] Similarity is calculated based on two core dimensions: error type and error cause. If both dimensions match, the similarity is 100%; if only one dimension matches, the similarity is 50%.

[0037] Step 1: Convert the "error type and error cause" of each genuine error data into standardized feature items. For example, the error type only includes two categories: spelling and grammar. The error cause is based on "error type + error location" and is further refined (e.g., "first letter confusion, last letter ed confusion, letter order reversal" under spelling errors, "predicate tense confusion, preposition collocation error, singular / plural confusion" under grammar errors). Step 2: Randomly select one unclassified true error data as the baseline error, and perform a two-dimensional matching of the remaining unclassified data with the baseline error in terms of "error type + error cause", and count the number of matching dimensions; Step 3: If both dimensions match, the similarity is determined to be ≥80%, and the two categories are classified together; if only one dimension matches, the similarity is determined to be <80%, and the two categories are classified together.

[0038] Specific steps for performing clustering operations: Step 1: Randomly select one unclassified true error from the population true error dataset as the baseline error, and label its error type, error cause, and core features; Step 2: Traverse the remaining unclassified true error data, and perform two-dimensional similarity matching with the baseline error one by one. Error data with similarity ≥ 80% are classified into the same category as the baseline error. Step 3: Select the next unclassified true error data as the new baseline error, and repeat Step 2 until all data in the population true error dataset have been clustered and classified. Step 4: Assign a unique category code to each type of error (e.g., 01-spelling error-ed ending confusion, 02-grammar error-predicate tense confusion), and label the core features of each type of error (error type, error cause, and core knowledge points involved).

[0039] For each type of error after clustering, frequency statistics were performed from both individual and group dimensions. All statistical indicators were objective and quantitative. Specific statistical content: Individual dimension: Statistical analysis of the number of times each type of error occurs and its frequency (number of errors of this type divided by the total number of true errors for this user), with annotations based on the user's anonymous ID; Group dimension: Statistical analysis of the total number of occurrences of each type of error in the group (the total number of occurrences of this type of error among all users), the group occurrence rate (the number of users who have this type of error divided by the total number of users in the group), and the group ranking (sorted by the group occurrence rate from high to low), with associated class or group identifiers and task ID labels.

[0040] The clustering and classification results are structurally integrated with the two-dimensional frequency statistics results (individual user dimension and class or group dimension) to form group misclassification feature data. The data structure is standardized and can be directly connected to the subsequent association strength identification process. Specific integration requirements: Each type of error is an independent feature data (which can be assigned a unique error category code), containing 5 modules: category code, core feature annotation, individual frequency statistics, group frequency statistics, and association identifier (class / group / task ID); Sort all categories by their occurrence rate from high to low to facilitate the rapid identification of high-frequency common errors in groups.

[0041] S203, based on the group error classification feature data, calculate the correlation strength value between various types of errors, construct an error correlation network, and structurally integrate the group error classification feature data and the error correlation network to obtain the group error map that represents the common error features and the error correlation features of the group.

[0042] Specifically, taking the group error classification feature data output by S202 as input, the correlation strength value between different error categories is first calculated from the individual user dimension to quantify the degree of correlation between errors; then, a visual error correlation network is constructed based on the correlation strength value to intuitively present the correlation logic between errors; finally, the group error classification features (including error type, cause, and frequency) and the error correlation network (including correlation strength) are structurally integrated to form a complete representation of the common error features of the group and the intrinsic correlation between errors.

[0043] Calculation of correlation strength between error categories: Co-occurring users: Users who make two or more types of errors in the same dictation task. Association strength value: The value ranges from 0 to 100%. The higher the value, the stronger the association between the two types of errors.

[0044] Calculation formula: Association strength value (error A and error B) = (Number of users co-occurring with both error A and error B ÷ Total number of users with either error A or error B) × 100%. Example: If there are 20 users in a class with a "spelling error ending in -ed" and 15 users with a "simple past tense error", and 10 users co-occurring with both types of errors, then the association strength value of the two types of errors = (10 ÷ (20 + 15 - 10)) × 100% = 40%.

[0045] To facilitate subsequent network construction and analysis, three levels are defined based on the association strength value, with each level labeled as a core feature: Strong association: Association strength value ≥ 30%; Medium correlation: correlation strength value 10%-30% (excluding); Weak association: Association strength value <10% (weak associations are only recorded and not included in the association network).

[0046] Based on the calculated association strength values, a visualized undirected network is constructed (without a clear causal direction, only showing the association relationship). The network structure is simple and intuitive, conforming to the viewing habits of instructors. 1. Definition of core network elements Nodes: Each clustered error category (e.g., "01-Spelling error-ed ending confusion" "02-Syntax error-Simple past tense confusion"). The size of a node is positively correlated with the "group occurrence rate" of that error category (the higher the occurrence rate, the larger the node). Edge: A line segment connecting two error nodes, representing a correlation between the two types of errors. The thickness of the edge is positively correlated with the "correlation strength value" (the higher the strength, the thicker the edge). Edge labeling: Directly label the association strength value (e.g., 40%) and association level (e.g., "strong association").

[0047] 2. Network Construction Rules Only retain erroneous connections of "strong association" and "medium association", and eliminate weak associations to avoid network redundancy; The network is constructed separately according to the dimension of "class or group".

[0048] The group error classification feature data and error association network are integrated in a two-layer structure to form a complete group error map, which includes two parts: a structured data table and a visualized association network. The error category code is used as the unique association identifier for both parts.

[0049] Based on the above error association network, the association strength values ​​between various types of errors are filtered, and error categories that meet the preset association strength are combined sequentially according to the logical relationship between knowledge points to form an error association combination sequence, which is used for matching and generating subsequent review materials.

[0050] For example, the system detected: Error A: The ending "ed" is misspelled; Error B: The simple past tense is grammatically incorrect; Error A and error B are strongly correlated; Knowledge point logic: Learn spelling first, then learn grammar; The system automatically combines the errors into: Error Association Combination Sequence = Error A to Error B.

[0051] S300, quantify and visualize the common error features and error correlation features in the group error map, and generate a teaching insight dashboard with standardized group learning data.

[0052] First, the qualitative common error characteristics and error correlation characteristics in the group error map are transformed into calculable numerical indicators. Then, the two types of numerical characteristics are subjected to dimensional unification and data normalization to eliminate differences in data units and form structured and standardized group learning data. Finally, the standardized data is mapped into multi-dimensional visualization charts that can be intuitively interpreted by teachers and integrated into a teaching insight dashboard to support teaching decisions (such as the explanation of key knowledge points and the generation of personalized review materials).

[0053] Specifically, generating a teaching insight dashboard with standardized group learning data may include the following steps: S301, the common error features and error correlation features in the group error map are numerically quantized and extracted to obtain the corresponding quantitative feature set of common errors and quantitative feature set of error correlation.

[0054] Specifically, all the common error features and error correlation features identified in the group error map are converted into fixed numerical indicators. The conversion process relies on the objective data that has been statistically completed in S202 and S203.

[0055] For the numerical extraction of common error features of a group, the error category code is used as the index to extract calculable numerical content corresponding to each type of error from the group error map, including the group occurrence rate, the total number of times the group occurs, the group ranking, and the average frequency of occurrence of an individual. The average frequency of occurrence of an individual is obtained by dividing the total number of times the group has this type of error by the number of users who have this type of error. All the extracted numerical content is combined to form a quantitative feature set of common errors of the group.

[0056] For the numerical extraction of error association features, the error category coding pairs are used as the index to extract the computable numerical content corresponding to each group of associated errors from the group error map, including the association strength value and the number of co-occurring users. At the same time, the association level is converted into the corresponding numerical identifier. All the extracted numerical content is combined to form a quantitative feature set of error association.

[0057] Using the same method as S203 for calculating the correlation strength value, the correlation strength value is obtained by dividing the number of users who simultaneously experience both types of errors by the total number of users who experience either Type I or Type II errors, and then multiplying by 100%. This value is used to intuitively reflect the degree of correlation between the two types of errors.

[0058] S302, the group common error quantitative feature set and error correlation quantitative feature set are integrated by dimension normalization and data normalization to obtain structured and standardized group learning data.

[0059] By unifying data dimensions and eliminating differences in units, the two types of quantitative feature sets are integrated into structured data that can be directly used for visualization. All operations are performed using conventional data processing methods.

[0060] The specific implementation method of dimension normalization is as follows: the association dimensions of the group common error quantification feature set and the error correlation quantification feature set are unified as class or group identifier, task ID, and error category code, to ensure that the two types of quantification feature sets can be accurately matched through the above unified dimensions. The values ​​and corresponding relationships of all dimensions are based on the content determined in S100 to S203.

[0061] The specific implementation of data normalization is as follows: using the basic principle of linear normalization, the total number of occurrences of the group, the average frequency of occurrence of individuals, and the number of co-occurring users in the error correlation quantitative feature set of the group common error are uniformly scaled to the numerical range of zero to one, eliminating the difference in the units of measurement between different numerical indicators and ensuring the rationality of subsequent visualization.

[0062] The specific implementation method of data integration is as follows: the quantitative feature set of common group errors and the quantitative feature set of error correlation after dimensional normalization and data standardization are merged with class or group identifier, task ID, and error category code as the core correlation basis. All numerical contents of the two types of quantitative feature sets are integrated into a single structured data. This structured data is the standardized group learning data, which can be directly used for subsequent visualization output.

[0063] S303, the structured and standardized group learning data is visualized and output in multiple dimensions to generate a teaching insight dashboard that carries the standardized group learning data.

[0064] Specifically, standardized group learning data is mapped into conventional visualization content according to the actual needs of teaching analysis, and all visualization content is then integrated into a unified teaching insight dashboard to meet the feasibility requirements of basic teaching analysis.

[0065] Standardized data corresponding to common error characteristics of a group can be mapped into a horizontal bar chart. Error category codes serve as the basis for chart identification, while corresponding error types and causes are labeled. The group occurrence rate serves as the numerical basis for the chart. The length of the bar chart visually displays the group occurrence rate of each type of error, and the high-frequency errors ranking at the top of the group are distinguished by the chart identification.

[0066] The standardized data corresponding to the error association features can be mapped into an undirected association network visualization. Each error category code is used as a node in the network. The size of the node is positively correlated with the total number of occurrences of the normalized group. The connection lines between nodes are positively correlated with the normalized association strength value. The connection lines are simultaneously labeled with the corresponding association strength value and association level. Only the error category code pairs with strong and moderate association are displayed, intuitively presenting the association relationship between different errors.

[0067] At the top of the teaching insight dashboard, a core indicator display area is set up to show the current class or group identifier, the total number of true errors in the dictation task corresponding to the current task ID, the error category code with the highest group occurrence rate and the corresponding error cause, the highest correlation strength value, and the number of strongly correlated error category code pairs. The core learning situation conclusions are presented intuitively through numerical data.

[0068] At the bottom of the teaching insight dashboard, there is an area for comparing individual learning progress. Using anonymous user IDs as identifiers, it displays the error category code for each user, the number of errors made by that user, and the average frequency of errors made by individuals at the group level, intuitively presenting the difference in error distribution between individual users and the group as a whole.

[0069] The above-mentioned horizontal bar visualization, undirected network visualization, core indicator display area, and individual learning comparison display area are integrated into one to form a complete teaching insight dashboard. This dashboard carries all standardized group learning data and can be directly used by teachers to view and analyze group learning.

[0070] S400: Taking the standardized group learning data from the teaching insight dashboard as input, the data is processed through a preset teaching material generation algorithm to obtain a group review material package and target learning quantitative indicators that match the group review material package.

[0071] Using the standardized group learning data output by S302 as input data, the common error quantitative features and error correlation quantitative features of the group are first extracted from the data to determine the weak points parameters of the corresponding group's dictation learning. Then, the weak point parameters are substituted into the preset teaching material generation algorithm, and based on the existing dictation teaching knowledge point system and teaching material resource library, a group review material package adapted to the group's learning is generated. Finally, combining the training objectives of the group review material package with the current quantitative level of the group's dictation learning, multi-dimensional target learning quantitative indicators are set to form a complete and matched target learning quantitative indicator system, providing a quantitative basis for evaluating the group's dictation review effect.

[0072] Specifically, obtaining the group review material package and the target learning progress quantitative indicators matching the group review material package may include the following steps: S401, extract the quantitative features of common group errors and the quantitative features of error correlation from the standardized group learning data of the teaching insight dashboard, and determine the core weak point parameters of the group's dictation learning.

[0073] Specifically, the quantitative characteristics of common group errors and the quantitative characteristics of error correlations are extracted from the standardized group learning data, which serve as the basis for subsequent matching of teaching materials.

[0074] From standardized group learning data, common error quantitative features are extracted, including group occurrence rate, total number of occurrences, group ranking, and average frequency of occurrence per individual. Error correlation quantitative features are also extracted, including correlation strength, correlation level, and number of co-occurring users. Based on a ranking rule from highest to lowest group occurrence rate, the common error quantitative features corresponding to the top-ranked error category codes are selected (the number of top-ranked selections can be set as needed, and this application does not limit this). Then, combining the correlation strength and correlation level, error correlation quantitative features corresponding to error category codes with strong or moderate correlations with these error category codes are selected. These top-ranked common error quantitative features are integrated and used as the core weakness parameters of the group's dictation learning, serving as a direct basis for matching subsequent teaching materials.

[0075] S402, input the core weak points parameters of the group's dictation learning situation into the preset teaching material generation algorithm, and combine the dictation teaching knowledge point system and teaching material resource library to generate a group review material package that is adapted to the group's learning situation.

[0076] Specifically, the quantitative features obtained from step S401 are used as the matching criteria. The pre-set teaching material generation algorithm is used to perform rule matching, retrieve corresponding materials from the dictation teaching knowledge point system and teaching material resource library, and integrate them to form a group review material package.

[0077] After inputting the core weaknesses of the group's dictation learning into a preset teaching material generation algorithm, the algorithm matches the error category code, error type, error cause, and corresponding dictation knowledge point in the quantitative features with the dictation teaching knowledge point system to determine the scope of review knowledge points to be covered. Based on the matched scope of review knowledge points, the algorithm retrieves corresponding dictation exercises, knowledge point explanations, and error correction examples from the teaching material resource library; it combines materials corresponding to the same error category code and integrates materials corresponding to related error category codes, ultimately generating a group review material package adapted to the group's learning situation. Alternatively, the core weaknesses of the group's dictation learning can be input into a preset teaching material generation algorithm. Combining the dictation teaching knowledge point system with the teaching material resource library, the algorithm sequentially retrieves teaching materials for corresponding knowledge points according to the error association combination sequence, generating a group review material package adapted to the error association sequence. This ensures that the review materials sequentially cover the knowledge points corresponding to the associated errors, improving the adaptability and coherence of the group review material package. For example, materials are generated in sequence from error A to error B: first, dictation exercises for words ending in -ed; then, dictation exercises for words in the simple past tense; these are pushed out in order and appear in sets according to their related order.

[0078] For example, the quantitative features of common group errors and the quantitative features of error correlation are input into a preset teaching material generation algorithm (the algorithm can be a matching algorithm, such as a preset tag matching rule, which performs a "one-to-one / many-to-many" fuzzy match between the input quantitative features (error category code, error cause, knowledge point) and the tags (knowledge point, error type) of the materials in the material library). The algorithm then matches the error category code 01, error type spelling error, error cause spelling confusion of words ending in -ed, and corresponding dictation knowledge point regular verb past tense spelling, and error category code 02, error type grammatical error, error cause simple past tense verb usage error, and corresponding dictation knowledge point simple past tense verb usage with the dictation teaching knowledge point system to determine the scope of review knowledge points that need to be covered as regular verb past tense spelling and simple past tense verb usage.

[0079] Based on the range of review knowledge points matched, the algorithm retrieves corresponding past tense dictation exercises for regular verbs, past tense predicate verb dictation exercises, explanations of spelling of verbs ending in -ed, explanations of the usage of the past tense, examples of spelling errors ending in -ed, and examples of grammatical errors in the past tense from the teaching material resource library. It then combines the exercises, explanations, and examples corresponding to error category code 01, and combines the exercises, explanations, and examples corresponding to error category code 02. Finally, it integrates the materials corresponding to related error category codes 01 and 02 to generate a group review material package.

[0080] It should be noted that the logical relationship between the pre-set teaching material generation algorithm, the dictation teaching knowledge point system, and the teaching material resource database is as follows: Dictation teaching knowledge point system: Provides matching rules and generation standards for the pre-set teaching material generation algorithm, determines what knowledge points and difficulty of exercises to generate, as well as the standard answers and error-prone annotation specifications; Teaching Material Resource Library: Provides ready-made materials (words, sentences, answers, annotations) that can be directly accessed by the preset teaching material generation algorithm; The pre-defined teaching material generation algorithm serves as the core of the process, matching content from the teaching material resource library according to the standards of the knowledge point system.

[0081] In addition, in actual work, the preset teaching material generation algorithm may fail to match certain situations. Therefore, exercises, answers, and error-prone annotations can be automatically generated according to the standards of the dictation teaching knowledge point system. The specific solution is as follows: The teaching material generation algorithm checks the matching results in the material library. If no matching material is found or the matching degree is insufficient (similarity is less than a preset threshold), the large model execution process is initiated. The teaching material generation algorithm assigns the core weakness parameters of the group's dictation learning to the learning adaptation dimension parameters of the large model, thus completing the model input initialization; The teaching material generation algorithm retrieves the dictation teaching knowledge point system, inputs the scope of review knowledge points into the large model, and constructs the boundary of the large model; The teaching material generation algorithm sets the question type structure, quantity, and relevance of the large model to obtain dictation practice questions and corresponding answers.

[0082] S403, based on the training objectives of the group review material package and the current quantitative level of the group's dictation learning, set multi-dimensional quantitative indicators of the target learning, and form a quantitative indicator system of the target learning that matches the group review material package.

[0083] Specifically, based on the current quantitative characteristic values ​​of the group, and combined with the training objectives of the group's review materials package, expected values ​​are set for the indicators corresponding to the quantitative characteristics, forming a target learning quantitative indicator system that can be used to evaluate the review effect.

[0084] The current quantitative characteristic values ​​of the group are calculated using the group occurrence rate and average individual occurrence frequency from the group's common error quantitative characteristics, and the correlation strength value from the error correlation quantitative characteristics. Based on the training objectives of the group review material package, target values ​​for corresponding indicators are set: the first category is the target group occurrence rate, the second category is the target individual average occurrence frequency, and the third category is the target correlation strength value. These target values ​​are then linked to the corresponding error category codes, error types, and error causes to form a multi-dimensional quantitative indicator system for target learning. This indicator system is fully matched to the content of the group review material package and can be directly used for the quantitative evaluation of subsequent group dictation review effectiveness.

[0085] S500, apply the group review material package to the group users in the class or group, obtain the secondary learning behavior data of the users after use, and analyze the secondary learning behavior data to obtain the result learning quantitative indicators.

[0086] Push the group review material package to the corresponding class / group users, organize targeted dictation review, and collect secondary dictation learning behavior data generated by users during the review process; extract secondary dictation feature data consistent with the quantitative feature dimensions mentioned above from this data; perform quantitative statistics and analysis on the group common error features and error correlation features of the secondary dictation feature data, and calculate the final learning quality quantitative indicators reflecting the review effect by combining the target learning quality quantitative indicators set in S403.

[0087] Specifically, obtaining the quantitative indicators of learning progress can include the following steps: S501, acquire secondary dictation learning behavior data generated during the user's review process.

[0088] Among them, the secondary dictation learning behavior data refers to the data generated during the user's review process when the group review material package is pushed to group users in the class or group and targeted dictation review is organized.

[0089] Specifically, through standardized dictation review data collection methods, we obtain full behavioral data generated by users after completing targeted dictation review using group review material packages. The data dimensions completely correspond to the error characteristics and quantitative indicator dimensions mentioned above.

[0090] For example, a group review material package (containing exercises / explanations / error correction examples for the past tense spelling of regular verbs and the usage of the simple past tense predicate) is pushed to the group users (anonymous IDs: U000001-U000040) of Class 1, Grade 3, and the class is organized to conduct a 20-minute targeted dictation review. The full data of the users' completion of the review exercises is collected through the dictation answering system. The specific data on secondary dictation learning behavior includes: Basic identifiers: anonymous user ID (U000001 / U000002…), class identifier (Grade 3, Class 1), task ID (review task T2026032002); Error-related data: Error category code (01 / 02) in user answers, error type (spelling error / grammar error), error cause (spelling confusion at the end of -ed / incorrect use of verb in the simple past tense), and the number of times this type of error occurred per user.

[0091] S502, extract features from the secondary dictation learning behavior data to obtain secondary dictation feature data.

[0092] Specifically, from the secondary dictation learning behavior data, data related to common group error features and error correlation features are screened and extracted to form secondary dictation feature data. The extraction dimensions are completely consistent with the S301 quantitative feature dimensions. Only data screening is performed, and no new features are added.

[0093] S503, the common error features and error correlation features of the group in the secondary dictation feature data are quantitatively extracted and analyzed, and the results of the quantitative indicators of learning situation are calculated in combination with the corresponding dimensions of the target learning situation indicators.

[0094] Specifically, statistical quantification is performed on the secondary dictation feature data at the group level to obtain the quantitative features of common group errors and the quantitative features of error correlation after review; this value is compared with the target learning quality quantitative indicators set in S403 and the original quantitative values ​​before review to calculate the result learning quality quantitative indicators reflecting the review effect, which can be calculated using basic statistics and difference / ratio calculation.

[0095] For example, the difference rate = (value before review - value after review) ÷ value before review × 100%, and the association strength value is calculated using the S203 formula.

[0096] Error category code 01: Group occurrence rate target value 30%, actual value 25%, difference rate 64.3%; average occurrence frequency per individual target value 2.5 times, actual value 2 times, difference rate 52.9%; Error category code 02: Group occurrence rate target value 15%, actual value 12.5%, difference rate 80.8%; Association strength between error category codes 01 and 02: Target value 30%, actual value 25%, difference rate 37.5%.

[0097] The actual quantitative values ​​of each dimension are all lower than the target values, forming the corresponding quantitative indicators of the class's learning situation. The difference rate is only used to identify the direction of optimization.

[0098] S600, based on the comparison difference between the result learning quality quantitative index and the target learning quality quantitative index, iteratively optimize the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for association strength identification, and the logic of the preset teaching material generation algorithm, so as to achieve targeted generation of dictation error pattern analysis and review materials for group users within a class or group.

[0099] Based on the comparison between the quantitative indicators of the learning situation and the target quantitative indicators of the learning situation, the temporary error identification rules, cluster similarity classification thresholds, association strength identification criteria, and teaching material generation algorithm logic are iteratively optimized to improve the accuracy of dictation error pattern analysis, enhance the targeting of review material generation, and provide learning support that is adapted to the actual learning situation for group users in classes or groups.

[0100] Specifically, achieving dictation error pattern analysis and targeted generation of review materials for group users within a class or group may include the following steps: S601, calculate the difference between the resulting quantitative indicators of learning and the target quantitative indicators of learning in the corresponding dimensions, and generate multi-dimensional quantitative analysis results of learning differences.

[0101] By calculating the difference rate of each indicator dimension using a unified formula, an intuitive quantitative analysis result of learning differences is formed, clarifying the direction for subsequent rule and algorithm optimization.

[0102] For example, the difference rate = (|target value - actual value after review| ÷ target value) × 100%; Based on the actual learning situation of the class, the occurrence rate of error category code 01, the average occurrence frequency of 01 individuals, the occurrence rate of 02 group, and the difference rate of the correlation strength value between 01 and 02 were calculated to form a multi-dimensional quantitative analysis result of learning differences, and to identify the optimization space of error analysis and material matching.

[0103] S602, based on the quantitative analysis results of the learning differences, the parameters of the rules for identifying temporary typos, the classification threshold for cluster similarity classification, and the judgment criteria for identifying association strength are optimized and iterated. At the same time, the feature weights of the logic of the preset teaching material generation algorithm are adjusted to obtain the adjustment parameters.

[0104] Specifically, based on the quantitative analysis results of learning differences, the various analysis rules and algorithm logic are adapted and optimized to form adjustment parameters that fit the learning situation of the class.

[0105] For example, the core basis of the iterative temporary typographical error identification rule is the individual average frequency difference rate of S601, combined with the actual error data of the class, to determine that in a single error, there is a real knowledge-based error accompanied by a related error. The original judgment condition cannot distinguish between typographical errors and real errors. Optimization: Based on the original single error judgment, a related error is added as a judgment dimension to improve the accuracy of distinguishing between real errors and occasional mistakes.

[0106] The core basis for adjusting the clustering similarity classification threshold is the error pattern correlation difference rate of S601. Combined with the actual error characteristics of the class, it was determined that the feature similarity of a large number of dictation errors within the class is concentrated in the 75%-79% range. The original 80% threshold could not complete the grouping of similar errors. Optimization: The classification threshold was adjusted from 80% to 75%, improving the range of similar error patterns and enhancing the completeness of error pattern recognition.

[0107] The core basis for optimizing the correlation strength identification criteria is as follows: S601 clarifies that the actual correlation strength value between error categories 01 and 02 in the class is 25%. This value represents the true correlation characteristics of errors in the class, and the original strong correlation threshold of 30% does not match this actual value. The optimization involves adjusting the strong correlation threshold to 25%, the medium correlation range to 10%-24%, and the weak correlation threshold to below 10%, thus aligning the criteria with the actual correlation characteristics of errors in the class.

[0108] The core basis for adjusting the feature weights of the teaching material generation algorithm is the difference rate of the group occurrence rate of S601. Combined with the actual distribution of error causes in the class, the degree of matching between review materials and core error causes is determined, which is a key factor affecting learning indicators. The original weights did not adequately reflect the error causes. Optimization: Increase the feature weights of error causes, strengthen their reference role in material matching, and improve the matching degree between review materials and core weaknesses.

[0109] The above adjustments are all based on the quantitative difference data of S601 and the statistical characteristics of actual errors in the classes as the core reasons, forming adjustment parameters that are adapted to the current learning situation, and providing a basis for subsequent rule updates.

[0110] S603, using the adjustment parameters, update the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for association strength identification, and the logic of the preset teaching material generation algorithm, to achieve targeted generation of dictation error patterns and review materials for group users within a class or group.

[0111] Specifically, the adjustment parameters obtained in S602 are synchronously updated into the full-process rules for temporary typographical error identification, cluster similarity classification, association strength identification, and teaching material generation.

[0112] The updated rules and algorithms are applied to the next round of dictation error pattern analysis and review material generation for classes or groups. They can accurately identify the real error characteristics of group users and generate review materials that fit the core weaknesses, achieving the goal of accurate analysis of dictation error patterns and targeted generation of review materials, and providing users with targeted learning support.

[0113] Simultaneously, based on the review effectiveness corresponding to the incorrect association combination sequences, the association strength judgment criteria and incorrect association combination rules are iteratively updated to make the generated incorrect association combinations and review materials more closely match the group's dictation learning situation, further realizing the targeted generation of review materials. For example, if students still make many mistakes after completing the A to B sequence, the system automatically adjusts the association strength threshold and combination order to generate a more suitable sequence for the next learning situation.

[0114] Secondly, this application provides a collaborative dictation and group error pattern analysis system, applied to the aforementioned collaborative dictation and group error pattern analysis method, comprising: The acquisition unit is used to acquire learning behavior data of a group of users within a class or group, and to extract features to obtain a group dictation feature dataset. The processing unit is used to perform temporary error identification, cluster similarity classification, and association strength identification on the data in the group dictation feature dataset in sequence to obtain a group error map, wherein the group error map represents the common error features and error association features of the group; The quantization unit is used to quantify and visualize the common error features and error correlation features in the group error map, and generate a teaching insight dashboard with standardized group learning data. The generation unit is used to take the standardized group learning data of the teaching insight dashboard as input, process it through a preset teaching material generation algorithm, and obtain a group review material package and a target learning quantitative indicator that matches the group review material package. The verification unit is used to apply the group review material package to the group users in the class or group, obtain the secondary learning behavior data of the users after use, and analyze the secondary learning behavior data to obtain the result learning quantitative indicators. The results unit is used to iteratively optimize the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for association strength identification, and the logic of the preset teaching material generation algorithm based on the comparison difference between the result learning quantitative indicators and the target learning quantitative indicators, so as to achieve targeted generation of dictation error pattern analysis and review materials for group users within a class or group.

[0115] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the aforementioned collaborative dictation and group error pattern analysis method.

[0116] Fourthly, this application provides a computer program that, when executed by a processor, implements the steps of the aforementioned collaborative dictation and group error pattern analysis method.

[0117] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0118] Obviously, those skilled in the art should understand that the various units or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0119] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A collaborative dictation and group error pattern analysis method, characterized in that, include: Acquire learning behavior data of a group of users within a class or group, and extract features to obtain a group dictation feature dataset; The data in the group dictation feature dataset are subjected to temporary error identification, cluster similarity classification, and association strength identification operations in sequence to obtain a group error map, wherein the group error map represents the common error features and error association features of the group; The common error features and error correlation features of the group in the group error map are quantified and visualized to generate a teaching insight dashboard with standardized group learning data; Using the standardized group learning data from the teaching insight dashboard as input, the data is processed through a preset teaching material generation algorithm to obtain a group review material package and target learning quantitative indicators that match the group review material package. The group review material package is applied to the group users in the class or group, and the secondary learning behavior data of the users after use is obtained. The secondary learning behavior data is analyzed to obtain the result of quantitative indicators of learning progress. Based on the comparison between the quantitative indicators of the learning situation and the quantitative indicators of the target learning situation, the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for the identification of association strength, and the logic of the preset teaching material generation algorithm are iteratively optimized to achieve targeted generation of dictation error patterns and review materials for group users within a class or group.

2. The collaborative dictation and group error pattern analysis method as described in claim 1, characterized in that, The steps of acquiring learning behavior data of a group of users within a class or group, and extracting features to obtain a group dictation feature dataset include: Obtain dictation-related learning behavior data from group users within a class or group; The dictation learning behavior data is preprocessed by sequentially performing data cleaning, outlier removal, and normalization to obtain preprocessed data. The dictation error features, answering behavior features, and common features of group answers are extracted and structured from the preprocessed data to obtain a group dictation feature dataset.

3. The collaborative dictation and group error pattern analysis method as described in claim 1, characterized in that, The step of performing temporary error identification, cluster similarity classification, and association strength identification on the data in the group dictation feature dataset to obtain a group error map includes: Based on the preset rules for identifying typos, the error data in the group dictation feature dataset is identified, and the false error data corresponding to temporary typos are removed to obtain the group true error dataset. A clustering similarity judgment threshold is set, and the similarity is calculated and clustered for the true error dataset of the group according to error type and error cause. The frequency of occurrence of each type of error in the group is counted to obtain the group error classification feature data. Based on the group error classification feature data, the correlation strength values ​​between various types of errors are calculated, and an error correlation network is constructed. The group error classification feature data and the error correlation network are then structurally integrated to obtain the group error map that represents the common error features and error correlation features of the group.

4. The collaborative dictation and group error pattern analysis method as described in claim 1, characterized in that, The step of quantifying and visualizing the common error features and error correlation features in the group error map to generate a teaching insight dashboard with standardized group learning data includes: The common error features and error correlation features in the group error map are numerically quantized and extracted to obtain the corresponding quantitative feature set of common errors and quantitative feature set of error correlation. The quantitative feature set of common errors and the quantitative feature set of error correlations of the group are integrated by dimension normalization and data normalization to obtain structured and standardized group learning data. The structured and standardized group learning data is visualized and output in multiple dimensions to generate a teaching insight dashboard that carries the standardized group learning data.

5. The collaborative dictation and group error pattern analysis method as described in claim 1, characterized in that, The step of taking standardized group learning data from the teaching insight dashboard as input, processing it through a preset teaching material generation algorithm, and obtaining a group review material package and target learning quantitative indicators matching the group review material package includes: From the standardized group learning data in the teaching insight dashboard, extract the quantitative features of common group errors and the quantitative features of error correlation to determine the core weak points parameters of the group's dictation learning. The core weaknesses of the group's dictation learning are input into the preset teaching material generation algorithm. Combined with the dictation teaching knowledge point system and teaching material resource library, a group review material package adapted to the group's learning situation is generated. Based on the training objectives of the group review materials package and the current quantitative level of the group's dictation learning, multi-dimensional quantitative indicators of learning progress are set to form a quantitative indicator system of learning progress that matches the group review materials package.

6. The collaborative dictation and group error pattern analysis method as described in claim 1, characterized in that, The steps of applying the group review material package to users within the class or group, obtaining secondary learning behavior data after the users' use, and analyzing the secondary learning behavior data to obtain quantitative indicators of learning progress include: Acquire secondary dictation learning behavior data generated during the user's review process. The secondary dictation learning behavior data refers to the data generated during the user's review process when the group review material package is pushed to group users in the class or group and targeted dictation review is organized. Feature extraction is performed on the secondary dictation learning behavior data to obtain secondary dictation feature data; The secondary dictation feature data is subjected to quantitative extraction and analysis of common group error features and error correlation features. Combined with the corresponding dimensions of the target learning quality quantitative indicators, the results of the learning quality quantitative indicators are obtained.

7. The collaborative dictation and group error pattern analysis method as described in claim 1, characterized in that, The steps of iteratively optimizing the rules for identifying temporary typos, the classification threshold for cluster similarity categorization, the judgment criteria for association strength identification, and the logic of the preset teaching material generation algorithm based on the comparison difference between the resulting quantitative indicators of learning progress and the target quantitative indicators of learning progress, to obtain the steps for achieving targeted generation of dictation error patterns and review materials for group users within a class or group, include: Calculate the difference between the resulting quantitative indicators of learning and the target quantitative indicators of learning in the corresponding dimensions, and generate multi-dimensional quantitative analysis results of learning differences; Based on the quantitative analysis results of the learning differences, the parameters of the rules for identifying temporary typos, the classification threshold for cluster similarity classification, and the judgment criteria for identifying association strength are optimized and iterated. At the same time, the feature weights of the logic of the preset teaching material generation algorithm are adjusted to obtain the adjustment parameters. Using the aforementioned adjustment parameters, the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for identifying association strength, and the logic of the preset teaching material generation algorithm are updated to achieve targeted generation of dictation error patterns and review materials for group users within a class or group.

8. A collaborative dictation and group error pattern analysis system, characterized in that, The collaborative dictation and group error pattern analysis method applied to any one of claims 1-7 includes: The acquisition unit is used to acquire learning behavior data of a group of users within a class or group, and to extract features to obtain a group dictation feature dataset. The processing unit is used to perform temporary error identification, cluster similarity classification, and association strength identification on the data in the group dictation feature dataset in sequence to obtain a group error map, wherein the group error map represents the common error features and error association features of the group; The quantization unit is used to quantify and visualize the common error features and error correlation features in the group error map, and generate a teaching insight dashboard with standardized group learning data. The generation unit is used to take the standardized group learning data of the teaching insight dashboard as input, process it through a preset teaching material generation algorithm, and obtain a group review material package and a target learning quantitative indicator that matches the group review material package. The verification unit is used to apply the group review material package to the group users in the class or group, obtain the secondary learning behavior data of the users after use, and analyze the secondary learning behavior data to obtain the result learning quantitative indicators. The results unit is used to iteratively optimize the rules for identifying temporary typos, the classification threshold for cluster similarity classification, the judgment criteria for association strength identification, and the logic of the preset teaching material generation algorithm based on the comparison difference between the result learning quantitative indicators and the target learning quantitative indicators, so as to achieve targeted generation of dictation error pattern analysis and review materials for group users within a class or group.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the collaborative dictation and group error pattern analysis method according to any one of claims 1 to 7.

10. A computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the collaborative dictation and group error pattern analysis method as described in any one of claims 1 to 7.

Images