Atomic ability training method and device for public examination preparation

Abstract

The application provides an atomization ability training method and device for public examination preparation, comprising: obtaining an ability training request containing a target ability identifier and a target level identifier; querying an ability training configuration table to obtain a corresponding atomic question type identifier, wherein the configuration table stores a mapping relationship among the ability identifier, the level identifier and the atomic question type identifier; extracting an independent atomic training question from an atomic question library according to the atomic question type identifier, wherein the independent training question is separated from a complete true question; sending the atomic training question to a user terminal and receiving answer data; associating and storing the answer data to an ability training record table; generating a learning report containing each ability dimension completion degree and accuracy rate according to the record table; identifying weak items based on the learning report and generating atomic training recommendation information for atomic ability training. The weak item identification and targeted recommendation based on the atomic ability decomposition, the data closed loop of learning, evaluation and improvement, and the multi-terminal unified learning portrait are realized.

Description

Technical Field

[0001] This invention relates to the field of online computer education technology, and in particular to an atomized ability training method and apparatus for civil service exam preparation. Background Technology

[0002] The civil service examination (including the written essay and structured interview) is the core component of selecting public officials. Its preparation involves complex skills development and extensive practice and feedback. Existing civil service exam preparation techniques mainly fall into the following categories: The first category is the traditional manual preparation method. Essay writing preparation relies on printed textbooks, in-person classes, and manual grading. Candidates practice with entire essays or sets of past exam questions, with teachers providing individual feedback. Interview preparation relies on in-person mock exams, with examiners or teachers providing on-site comments. While this method offers personalized feedback, manual grading is inefficient, has a long feedback cycle, is difficult to scale up to serve a large number of candidates, and the uneven distribution of teacher resources leads to high preparation costs.

[0003] The second category consists of general online question banks and essay grading tools. Currently, some platforms offer electronic question banks for test-takers to practice with, or use rule-based keyword matching and sentence structure detection to score essays; others use a single large model interface to provide overall scoring and general feedback. These tools are not specifically designed for the scoring criteria and grading habits of the essay writing section, making it difficult to cover the multi-dimensional evaluation requirements of essays, such as "thesis, structure, argumentation, and language." Furthermore, they do not break down the required essay writing skills into separately trainable units; training remains primarily focused on "whole questions and templates," lacking specificity.

[0004] The third category consists of interview simulation and speech recognition products. Existing interview preparation products primarily rely on question banks and mock questions, employing a "listen to the whole question—answer—compare and analyze" model. A few products incorporate speech recognition technology to track fluency or duration of responses. These products fail to break down the necessary interview skills (such as question identification, keyword extraction, contradiction recognition, intent judgment, fluency, and pacing) into independent, repeatedly trainable atomic modules, making it difficult for users to specifically strengthen their weaknesses.

[0005] The fourth category is multi-terminal distributed learning platforms. In existing solutions, learning data is scattered across multiple terminals or products such as mobile apps, web pages, and offline devices, making it impossible to form a unified learning profile and cross-terminal learning reports. The learning, practice, grading, writing, and reporting stages are independent of each other, and there is no closed-loop recommendation based on unified data to form a "learn-practice-evaluate-revise-practice" cycle.

[0006] Based on the above-mentioned existing technologies, the following obvious drawbacks exist: First, there is a lack of mechanisms for breaking down and specifically training atomic skills. Existing essay writing and interview preparation products all adopt the approach of "whole-question practice and full-set question drills," failing to break down the skills required for essay writing and interviews into atomic skill units that can be trained independently. The consequences are: users find it difficult to identify and address weaknesses in a single skill dimension (such as essay writing's question analysis, key point extraction, and logical sequencing, or interview's question identification, keyword extraction, and contradiction recognition); training lacks a clear progression path of "practicing atomic skills first, then comprehensive application"; and weaknesses are difficult for the system to identify and recommend corresponding atomic training, resulting in limited training relevance and preparation efficiency.

[0007] Second, the data in the learning process is fragmented and does not form a closed loop. In the existing solution, the learning materials, practice, correction, writing, and learning reports are independent of each other, and the data is not integrated, making it impossible to form a closed loop of "learning-practice-evaluation-correction-re-practice". The data is not consistent across multiple platforms, making it difficult to generate cross-platform learning reports and personalized recommendations, and it is impossible to form a unified profile of the user's learning behavior on different devices.

[0008] Third, reliance on a single large model leads to insufficient service stability. Existing review or writing assistance solutions mostly use a single large model interface. Once this service becomes unavailable or experiences rate limiting, the overall functionality is affected, users cannot obtain alternative capabilities, and the experience is poor.

[0009] Fourth, the grading standards are out of sync with the usual practice of grading essays for the civil service exam. General essay grading tools have not developed a structured grading process that includes reference points, scoring criteria, and score allocation for the essay question type. As a result, the scoring results are difficult to match with the "point-based scoring" practice of civil service exams, and they lack refined processing for matching key points and multi-dimensional scoring.

[0010] Fifth, content production and multi-platform distribution are disconnected. In the existing solution, the production, editing, and distribution of preparation content (such as news briefs, model articles, and hot topic materials) are scattered across different systems, resulting in inconsistent content across multiple platforms, high operating costs, and poor timeliness. Summary of the Invention

[0011] This invention provides an atomized ability training method and apparatus for civil service exam preparation, which addresses the shortcomings of existing technologies, such as the lack of atomized ability breakdown and specialized training mechanisms, fragmented data in the learning process without forming a closed loop, single-point dependence of review services, and a disconnect between review standards and essay marking habits. It achieves weakness identification and targeted recommendations based on atomized ability breakdown, a closed loop of learning, practice, review and revision data, intelligent review by multiple service providers, and a unified learning profile across multiple terminals.

[0012] This invention provides an atomized ability training method for civil service exam preparation, comprising: Obtain the user's ability training request in the civil service exam preparation system, wherein the ability training request includes a target ability identifier and a target level identifier; Based on the target ability identifier and the target level identifier, a preset ability training configuration table is queried to obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier. The ability training configuration table is used to store the mapping relationship between ability identifier, level identifier and atomic question type identifier. Based on the at least one atomic question type identifier, corresponding atomic training questions are extracted from the atomic question bank, wherein the atomic question bank stores independent training questions corresponding to each atomic question type identifier, and the independent training questions are separate from the complete real questions; The atomic training questions are sent to the user terminal, and the answer data returned by the user terminal is received. The answer data is associated with the target ability identifier, the target level identifier and the corresponding atomic question type identifier and stored in the ability training record table to obtain the atomic training record; Based on the atomic training records in the capability training record table, a learning report is generated, which includes the completion rate and accuracy rate for each capability dimension. Based on the learning report, weaknesses in ability are identified, and atomic training recommendation information is generated based on these weaknesses. Atomized ability training is then performed on the user based on the atomic training recommendation information.

[0013] In one possible implementation, the method further includes: The target competency identifiers include essay writing ability identifiers or interview workshop identifiers; When the target ability identifier is the essay writing ability identifier, the atomic question type identifier includes at least one of the following: text selection, fill-in-the-blank, logical sorting, key point classification, argument splicing, title argument, essay framework, oral to written language conversion, and text compression. When the target capability identifier is an interview workshop identifier, the method further includes: Based on the interview workshop identifier and the target level identifier, query the module configuration information to obtain at least one atomic module identifier corresponding to the target level identifier. The atomic module identifier corresponds to an independent training mode and question set.

[0014] In one possible implementation, the method further includes: The answer data is written into the ability training record table, which includes user identifier, question identifier, ability identifier, level identifier, atomic question type identifier, score or correct / incorrect identifier, and completion timestamp. The ability identifier is used to distinguish between the essay writing ability and the interview workshop, and the atomic question type identifier or atomic module identifier is used to distinguish the specific training format.

[0015] In one possible implementation, the method further includes: Receive a learning report generation request, the learning report generation request including user identifier and time range parameters; Filter the atomic training records that match the user identifier and time range parameters from the capability training record table; The system filters practice test records that match the user identifier and time range parameters from the practice test record table. The records in the practice test record table are answer data, test identifier, and completion timestamp that are written in real time in response to the user's historical practice test behavior. The practice test records and the atomic training records are stored in the same data storage. Based on the atomic training records and the set practice records, the accuracy rate under each ability dimension is calculated; Based on the accuracy rate under each capability dimension, a learning report is generated, which records radar chart data for each capability dimension.

[0016] In one possible implementation, the method further includes: Compare the accuracy rate for each capability dimension with preset thresholds; When the accuracy rate in a certain capability dimension is lower than the preset threshold, that capability dimension is determined to be a weak capability dimension. Based on the weak ability dimensions and their corresponding level identifiers, query the ability training configuration table to obtain recommended atomic question type identifiers; Based on the aforementioned weakness ability dimensions, level labels, and recommended atomic question type identifiers, atomic training recommendation information is generated.

[0017] In one possible implementation, the method further includes: Receive user-submitted test questions and answers; Write the answer data of the set of questions, the corresponding set of questions identifier, and the completion timestamp into the set of questions practice record table; The set of question answering data is sent to the review and scheduling module. The review and scheduling module maintains at least two AI review service instances, and each service instance is configured with a priority order. The review scheduling module calls AI review service instances sequentially according to the priority order. When a higher priority service instance is unavailable or returns an invalid result, it automatically switches to the next higher priority AI review service instance. Receive the review results, which include the total score, dimension score, key point matching list and comment information; The evaluation results are stored in the evaluation result table, and the evaluation results are associated with the corresponding records in the practice test record table.

[0018] In one possible implementation, the method further includes: Invoke the AI ​​review service instance with the current priority and send structured prompts, which include the question stem, material summary, reference points, scoring criteria and score allocation information; Receive response data and determine whether the response data meets the valid conditions. The valid conditions include that the response has not timed out, the response body can be parsed, the total score and the scores of each dimension are within the range of the question scores, and there are no missing required fields. When the response data does not meet the valid conditions, the current AI review service instance is marked as unavailable, and a call to the next priority AI review service instance is triggered. When all AI review service instances are unavailable, a default downgrade result or error message will be returned.

[0019] This invention also provides an atomized ability training device for civil service exam preparation, comprising the following modules: The acquisition module is used to acquire the user's ability training request in the civil service exam preparation system. The ability training request includes a target ability identifier and a target level identifier. The acquisition module is further configured to query a preset ability training configuration table based on the target ability identifier and the target level identifier, and obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier, wherein the ability training configuration table is used to store the mapping relationship between ability identifier, level identifier and atomic question type identifier; An extraction module is used to extract corresponding atomic training questions from an atomic question bank based on the at least one atomic question type identifier. The atomic question bank stores independent training questions corresponding to each atomic question type identifier, and the independent training questions are separate from the complete real questions. The sending and receiving module is used to send the atomic training questions to the user terminal and receive the answer data returned by the user terminal. The storage module is used to associate the answer data with the target ability identifier, the target level identifier and the corresponding atomic question type identifier and store them in the ability training record table to obtain the atomic training record; The generation module is used to generate a learning report based on the atomic training records in the capability training record table. The learning report includes the completion rate and accuracy rate under each capability dimension. The generation module is further configured to identify capability weaknesses based on the learning report, generate atomic training recommendation information based on the capability weaknesses, and perform atomic capability training on the user based on the atomic training recommendation information.

[0020] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the atomized ability training method for civil service exam preparation as described above.

[0021] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the atomic ability training method for civil service exam preparation as described above.

[0022] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the atomic ability training method for civil service exam preparation as described above.

[0023] The present invention provides an atomized ability training method and apparatus for civil service exam preparation. The method involves: acquiring a user's ability training request within a civil service exam preparation system, wherein the request includes a target ability identifier and a target level identifier; querying a preset ability training configuration table based on the target ability identifier and target level identifier to obtain at least one atomic question type identifier corresponding to the target ability identifier and target level identifier, wherein the ability training configuration table stores the mapping relationship between ability identifiers, level identifiers, and atomic question type identifiers; and extracting corresponding atomic training questions from an atomic question bank based on the at least one atomic question type identifier, wherein the atomic question bank stores information related to each atomic question type identifier. The system provides corresponding independent training questions, which are separate from the complete practice questions. These atomic training questions are sent to the user terminal, and the user terminal's response data is received. The response data is associated with the target ability identifier, target level identifier, and corresponding atomic question type identifier and stored in an ability training record table to obtain atomic training records. A learning report is generated based on the atomic training records in the ability training record table, and the learning report includes the completion rate and accuracy rate for each ability dimension. Based on the learning report, ability weaknesses are identified, and atomic training recommendation information is generated based on these weaknesses. Atomized ability training is then performed on the user based on the atomic training recommendation information. Compared to existing technologies that lack atomic capability breakdown and specialized training mechanisms, fragmented data in the learning process without forming a closed loop, single-point dependence on review services, and a disconnect between review standards and essay marking habits, this solution breaks down essay writing and interviews into atomic capability training and combines it with closed-loop learning, multi-service provider review, essay writing assistance, multi-terminal report generation and publishing. This enables weakness identification and targeted recommendations based on atomic capability breakdown, a closed-loop data system for learning, practice, review and revision, intelligent review by multiple service providers, and a unified learning profile across multiple terminals. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating the atomized ability training method for civil service exam preparation provided by the present invention.

[0026] Figure 2 This is a diagram of the atomization capability training system architecture provided by the present invention.

[0027] Figure 3 This is the overall architecture diagram of the atomized ability training system for civil service exam preparation provided by the present invention.

[0028] Figure 4 This is a flowchart of the essay writing and AI review process provided by this invention.

[0029] Figure 5 This is the AI ​​writing assistance flowchart provided by the present invention.

[0030] Figure 6 This is a flowchart of the content generation and publishing process for the management terminal provided by the present invention.

[0031] Figure 7 This is a schematic diagram of the structure of the atomized ability training device for civil service exam preparation provided by the present invention.

[0032] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0034] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0035] Figure 1 This is a flowchart illustrating the atomized ability training method for civil service exam preparation provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: S11. Obtain the user's skills training request in the civil service exam preparation system.

[0036] The skills training request includes the target skills identifier and the target level identifier.

[0037] In one embodiment, a user logs into the civil service exam preparation system via a mini-program or web terminal, and selects either "Essay Writing Skills Training" or "Interview Skills Advancement" on the homepage or skills training entry point. The user selects their target skill and level on the front-end interface, for example, "Key Points Summarization and Extraction Skills - Intermediate." The front-end encapsulates the user's selection into a skills training request and sends it to the back-end. The skills training request includes at least a user identifier, a target skill identifier, and a target level identifier. The target skill identifier distinguishes between different underlying essay writing skills (such as key points summarization and extraction, language compression and generalization, logical structure and organization, question analysis and thesis development, and argument and evidence organization) or different interview workshops (such as listening workshops, reading workshops, speaking workshops, and practice workshops). The target level identifier indicates the training difficulty level (such as beginner, intermediate, and advanced). S12. Based on the target ability identifier and target level identifier, query the preset ability training configuration table to obtain at least one atomic question type identifier corresponding to the target ability identifier and target level identifier.

[0038] The Ability Training Configuration Table is used to store the mapping relationship between ability identifiers, level identifiers, and atomic question type identifiers.

[0039] The ability training configuration table is pre-stored in the database to establish a mapping relationship between "ability + level → atomic question type". Taking the essay writing section as an example, when the target ability is identified as "logical structure and organization ability" and the target level is identified as "intermediate", querying the ability training configuration table will retrieve a list of atomic question type identifiers available for that ability and level, such as logical sorting and key point classification. Taking the interview section as an example, when the target ability is identified as "listening workshop" and the target level is identified as "L1 basic perception level", querying the module configuration information will retrieve a list of atomic module identifiers available for that level, such as question type identification, keyword extraction, and contradiction / task identification. The ability training configuration table is maintained by the operations or administrator, and the number of abilities, level tiers, and the correspondence between question types and abilities can all be configured and changed. New abilities or question types can be added without modifying the code.

[0040] S13. Based on the at least one atomic question type identifier, extract the corresponding atomic training questions from the atomic question bank.

[0041] The atomic question bank stores independent training questions corresponding to each atomic question type identifier. These independent training questions are separate from the complete real exam questions and are specifically for atomic ability training. In one embodiment, the backend, based on the atomic question type identifier obtained in step S12, randomly or sequentially selects a preset number (e.g., 10 questions) of atomic training questions from the atomic question bank according to the condition of "ability identifier + level identifier + atomic question type identifier". Each atomic training question includes a stem, options (if applicable), correct answer, explanation, associated ability identifier, level identifier, and atomic question type identifier. The atomic question bank and the full set / real exam question bank are independent of each other in terms of data model and storage, ensuring that atomic training questions can be added, deleted, modified, and queried independently of the complete real exam questions.

[0042] S14. Send the atomic training questions to the user terminal and receive the answer data returned by the user terminal.

[0043] The backend returns the extracted list of atomic training questions (including question content, question type identifiers, and the types of interactive components required for frontend rendering) to the user terminal via a REST API. The user terminal renders the questions according to the question type field, mapping it to different interactive components: for example, "Logical Sorting" questions are rendered as a drag-and-drop sorting interface, "Key Points Categorization" questions as a category drag-and-drop interface, "Fill in the Blanks" questions as an input box interface, and "Text Selection" questions as a clickable text highlighting interface. After the user answers each question, the frontend submits the answer data (including user identifier, question identifier, user answer, time taken, etc.) to the backend in batches or question by question.

[0044] S15. The answer data is associated with the target ability identifier, target level identifier and corresponding atomic question type identifier and stored in the ability training record table to obtain the atomic training record.

[0045] After receiving the answer data, the backend writes it to the Ability Training Record Table. The Ability Training Record Table must contain at least the following fields: User ID, Question ID, Ability ID, Level ID, Atomic Question Type ID (or Atomic Module ID for the Interview Side), Score or Correct / Incorrect ID, and Completion Timestamp. The Ability ID distinguishes between the essay writing skills and the interview skills, while the Atomic Question Type ID or Atomic Module ID distinguishes the specific training format. The Ability Training Record Table, along with the Practice Question Record Table, Review Result Table, and Writing Record Table, are stored in the same data store for easy aggregation and generation of a learning report later.

[0046] S16. Generate a learning report based on the atomic training records in the capability training record table.

[0047] In one embodiment, the learning report generation request is triggered by the user or periodically by the system. The request includes a user identifier and a time range parameter (e.g., the last 30 days). The backend filters atomic training records that match the user identifier and time range parameter from the ability training record table; simultaneously, it filters set practice records that match the same conditions from the set practice record table. The records in the set practice record table are answer data, set identifiers, and completion timestamps written in real time in response to the user's historical set practice behavior. Based on the atomic training records and the set practice records, the number of questions completed, accuracy rate, and score trend for each ability dimension are calculated; based on the accuracy rate for each ability dimension, ability dimension radar chart data is generated and assembled into a learning report, which is returned to the frontend. The learning report includes the completion rate and accuracy rate for each ability dimension.

[0048] S17. Identify the weaknesses in the learning report, generate atomic training recommendation information based on the weaknesses in the learning report, and perform atomic ability training on the user based on the atomic training recommendation information.

[0049] The accuracy rate for each ability dimension is compared with a preset threshold. When the accuracy rate for a certain ability dimension is lower than the preset threshold, that ability dimension is identified as a weak ability dimension. Based on the weak ability dimension and its corresponding level identifier, the ability training configuration table is queried to obtain the recommended atomic question type identifier. Based on the weak ability dimension, level identifier, and recommended atomic question type identifier, atomic training recommendation information is generated and sent to the user terminal.

[0050] The user terminal displays a recommendation card on the homepage or the capability advancement page. After the user clicks on it, they will enter the atomized capability training process from step S11 to step S15 again, forming an atomized capability training closed loop of "identifying weaknesses → recommending atomized training → executing training → updating records → reporting again".

[0051] The present invention provides an atomized ability training method for civil service exam preparation. This method involves: acquiring a user's ability training request from a civil service exam preparation system, where the request includes a target ability identifier and a target level identifier; querying a pre-defined ability training configuration table based on the target ability identifier and target level identifier to obtain at least one atomic question type identifier corresponding to the target ability identifier and target level identifier, wherein the ability training configuration table stores the mapping relationship between ability identifiers, level identifiers, and atomic question type identifiers; and extracting corresponding atomic training questions from an atomic question bank based on at least one atomic question type identifier, wherein the atomic question bank stores a list of atomic questions... The system generates independent training questions corresponding to each question type, separating these questions from the complete practice questions. Atomic training questions are sent to the user terminal, and the system receives the user's response data. The response data is then associated with the target ability identifier, target level identifier, and corresponding atomic question type identifier and stored in the ability training record table to obtain atomic training records. Based on the atomic training records in the ability training record table, a learning report is generated, containing the completion rate and accuracy for each ability dimension. Based on the learning report, ability weaknesses are identified, and atomic training recommendation information is generated based on these weaknesses. Atomized ability training is then conducted for the user based on the atomic training recommendation information. Compared to existing technologies that lack atomic capability breakdown and specialized training mechanisms, fragmented data in the learning process without forming a closed loop, single-point dependence of review services, and a disconnect between review standards and essay marking habits, this method breaks down essay writing and interviews into atomic capability training and combines it with closed-loop learning, multi-service provider review, essay writing assistance, multi-terminal report generation and publishing. This achieves weakness identification and targeted recommendations based on atomic capability breakdown, a closed loop of learning, practice, review and revision data, intelligent review by multiple service providers, and a unified learning profile across multiple terminals.

[0052] Figure 2 This is a diagram of the atomization capability training architecture provided by the present invention. For example... Figure 2 As shown, the atomized ability training system of the present invention includes an atomized training system for essay writing and an atomized training system for interviews.

[0053] On the essay writing side, the atomized ability training system adopts a three-tiered structure of "ability—level—atomic question type". The first level is the ability layer, defining several fundamental essay writing abilities, including but not limited to: the ability to summarize and extract key points, the ability to compress and summarize language, the ability to structure and organize logically, the ability to analyze the question and formulate a thesis, and the ability to organize arguments and evidence. The second level is the level layer, with multiple levels for each ability (e.g., beginner, intermediate, advanced), used to control the difficulty of questions and the depth of training. The third level is the atomic question type layer, defining various reusable atomic question types, including but not limited to: text selection, fill-in-the-blank, logical sequencing, key point classification, argument splicing, title and argument, essay framework, oral-to-written language conversion, and text compression. Atomic question types are atomic in nature and can be reused by multiple abilities. A mapping relationship of "ability identifier + level identifier → atomic question type identifier list" is established through an ability training configuration table. After a user selects a certain ability and level, the system determines the available atomic question types based on the mapping relationship and extracts corresponding questions from the atomic question bank. In addition, the essay section also includes specialized training layers (such as summarizing, comprehensive analysis, official document writing, and long essays), which break down complete past exam questions into stages to complement the basic skills training.

[0054] On the interview side, the atomized ability training system adopts a three-tiered structure: "Workshop—Level—Atomic Module". The first level is the workshop layer, including but not limited to: Listening Workshop (solving listening and question analysis), Reading Workshop (solving expression refinement), Speaking Workshop (solving viewpoint output), and Practice Workshop (providing real exam questions for practice). The second level is the level layer, such as L1 basic perception level, L2 advanced screening level, and L3 intensive adaptation level. The third level is the atomic module layer, with each level containing several atomic modules, each corresponding to an independent training format and question set. For example, the Listening Workshop L1 level may include atomic modules such as question type identification, keyword extraction, and contradiction / task identification; the Reading Workshop may include atomic modules such as sentence repetition, paragraph repetition, reading aloud without notes, rhythm alignment, and controlled-speed reading; the Speaking Workshop may include atomic modules such as viewpoint elaboration, strategy development, framework completion, and logical correction; and the Practice Workshop provides random real exam questions for practice. Each atomic module has its own set of questions and completion progress records. Users enter the module level by level, from workshop to level. After completing the questions in a module, the system updates the completion progress of that module and writes it into the ability training record table.

[0055] Figure 3 This is the overall architecture diagram of the atomized ability training system for civil service exam preparation provided by this invention. (See diagram below.) Figure 3 As shown, the system includes a user terminal, a backend service layer, a data storage layer, and an external service layer.

[0056] The user end includes both mini-program terminals and web terminals, both of which share the same backend REST API and identity authentication system (such as JWT). The user's login status on the mini-program or web corresponds to the same user identifier, ensuring that cross-platform behavioral data can be uniformly aggregated.

[0057] The backend service layer includes: Ability Training Service, Question Management Service, Learning Report Service, Recommendation Service, Review Scheduling Service, Writing Assistance Service, Content Generation Service, and User Account Service. The Ability Training Service handles ability training requests, queries the ability training configuration table, and manages the reading and writing of the atomic question bank and ability training record table. The Question Management Service maintains and extracts atomic question banks and practice / real exam question banks. The Learning Report Service aggregates data from the ability training record table, practice exam record table, review result table, and writing record table to generate learning reports and ability radar charts. The Recommendation Service generates atomic training recommendation information based on the weakness identification results in the learning reports. The Review Scheduling Service maintains multiple AI review service instances, calls them according to priority, and performs failover. The Writing Assistance Service constructs domain-specific prompts based on user requests and calls a large model to provide capabilities such as polishing, continuation writing, material adaptation, and full-text scoring. The Content Generation Service generates structured content based on template types and supports multi-platform publishing. The User Account Service manages unified identity authentication and permissions across multiple platforms.

[0058] The data storage layer includes a relational database or a document-oriented database, uniformly storing training record tables, practice test record tables, grading result tables, writing record tables, skills training configuration tables, atomic question banks, practice test and real test question banks, content libraries, and user account data. All behavioral data is associated with a unified user identifier and timestamp for easy cross-platform aggregation.

[0059] The external service layer includes multiple AI review service instances (such as service A, service B, etc.) and a large model interface. The review scheduling service calls each AI review service instance sequentially according to the configured priority order; the writing assistance service and content generation service obtain the generated results through the large model interface.

[0060] Figure 4 This is a flowchart of the essay writing and AI review process provided by this invention. Figure 4 As shown, the process includes submitting completed sets of questions, scheduling reviews from multiple service providers, constructing structured prompts, matching key points, and storing review results.

[0061] After completing a set of questions (such as a long essay or comprehensive analysis question) on the front end, the user submits the answer data. The back end receives the answer data and writes the answer data, the corresponding set identifier, and the completion timestamp into the set practice record table. Subsequently, the back end sends the answer data to the review and scheduling module.

[0062] The review scheduling module maintains at least two AI review service instances, each configured with a priority order. For each user submission, the review scheduling module calls the AI ​​review service instances sequentially according to priority. During the call, the review scheduling module constructs structured prompts and sends them to the current AI review service instance. The structured prompts include the question stem, material summary, reference points, scoring criteria, and score allocation information, and require the large model to return the total score, dimension scores, comments, and a list of matching points in a specified JSON structure.

[0063] After receiving the response data, the review and scheduling module determines whether the response data meets the validity conditions. Valid conditions include: the response has not timed out, the response body is parsable, the total score and scores for each dimension are within the range of the question's score, and no required fields are missing. When the response data meets the validity conditions, the module receives the review results, which include the total score, dimension scores, a list of matching key points, and commentary information. The review results are stored in the review results table and associated with the corresponding records in the practice test record table.

[0064] When the response data does not meet the valid conditions (such as network error, timeout, rate limiting, unparseable response body, score out of range, or missing required fields), the current AI review service instance is marked as unavailable, and the next lower priority AI review service instance is automatically invoked. When all AI review service instances are unavailable, a preset downgraded processing result (such as rule-based keyword matching score, or simply saving the answer data without returning AI comments) or an error message is returned.

[0065] In one embodiment, the key point matching can be further processed on the backend to match the user's answer with the reference key points, generating a list of "reference key points - user answer fragments - whether they match" for easy highlighting on the frontend. Dimensional scoring can be linked to the question type. For example, the essay can be scored using four dimensions: theme, structure, argumentation, and language, while the short answer questions can be scored using dimensions such as content and organization, to better align with the grading habits of civil service exams.

[0066] Figure 5 This is a flowchart of the AI-assisted writing process provided by this invention. For example... Figure 5 As shown, the process includes receiving writing assistance requests, constructing domain-specific prompts, calling large models, and parsing results in a structured manner.

[0067] Users submit writing assistance requests in the front-end writing interface. The writing assistance request includes an assistance type identifier, the topic context, and the user's input text. The assistance type identifier includes at least one of the following: polishing, continuation writing, material adaptation, and full text scoring.

[0068] The writing assistance service constructs domain-specific prompts based on the assistance type identifier. These prompts include constraints on the style of written essays, scoring criteria, and output format. For example, when the assistance type is continuation writing, the prompts explicitly require: no change to the candidate's original viewpoint and overall structure; better development around the scoring criteria; standard, concise, and elegant language; clear logic and natural transitions; output only the continuation sentences, not explanations. When the assistance type is polishing, the prompts require optimizing the language expression while maintaining the original meaning, making it conform to the standard of written essays. When the assistance type is full-text scoring, the prompts require scoring according to the essay grading standards from dimensions such as theme, structure, argumentation, and language, and providing paragraph annotations.

[0069] The writing assistance service sends domain-specific prompts and user-input text to a large model interface, and receives the writing assistance results returned by the large model. The service performs structured parsing of the writing assistance results, extracting dimension scores, paragraph annotations, or rewritten text, and returns it to the front end for display. Optionally, the writing assistance results can be stored in a writing record table for use in learning report aggregation.

[0070] Figure 6 This is a flowchart of the content generation and publishing process for the management terminal provided by this invention. For example... Figure 6 As shown, the process includes receiving content generation requests, generating structured content, storing content in the library, previewing and editing on multiple platforms, and publishing on multiple platforms.

[0071] Operations personnel select the content generation template type and generation parameters in the management backend and submit a content generation request. Template types include news briefs, daily quotes, essay samples, interview hot topics, and "Today in History," etc. Generation parameters include date, theme keywords, and section requirements.

[0072] The content generation service generates structured content data by calling the large model interface or built-in rules based on the template type identifier and generation parameters. The structured content data includes titles, body text, paragraphs, tags, and accompanying images, organized in JSON format. The generated result is written to the content library and assigned a unique content identifier; its initial state is a draft.

[0073] Operations personnel can preview the generated structured content in the management front-end interface and edit and adjust the main text, styles, and paragraphs. After confirmation, the management front-end renders the structured data into different export formats based on the template type: when the target is a WeChat Official Account publication, it is rendered as HTML code including the header, section title, item list, and styles. Operations personnel can copy the HTML to the WeChat Official Account backend or push it directly through the interface; when the target is a poster publication, a poster image is generated based on the structured data. The same content is stored in the database with the same content identifier. The user end (mini-program or web) retrieves and displays the content from the content database using this identifier or conditions such as content type and date. The management end and the user end share the same data source, realizing "one-time production, multi-terminal publication" and avoiding inconsistencies in content across multiple platforms.

[0074] This invention has the following technical advantages: (1) The essay writing and interview are broken down into atomic ability training, so that the training data (ability dimensions, levels, question types / modules) can be recorded and statistically analyzed by the system in a structured manner, thereby supporting data-based weakness identification and targeted recommendations, and improving the relevance and measurability of training and evaluation; and combined with comprehensive practice of full-set questions / real questions, a clear "atomic ability → comprehensive application" advancement path is formed. On the essay writing side, the configuration of "multiple abilities × multiple levels × multiple question types" allows different atomic question types to be selected for the same ability at different difficulty levels; on the interview side, the hierarchy of "workshop - level - atomic module" allows each training form to have corresponding questions and interactions, so that users can strengthen their weak links separately without having to do a full set of questions every time, achieving higher frequency and more focused training in a limited time, and receiving atomic module or full-set question recommendations generated by the system based on completion / accuracy / dimensional scores.

[0075] (2) A complete closed loop for essay preparation is formed. Data from each stage, including materials, flashcards, exercises, review, writing, reports, and recommendations, is stored in a unified manner and can be aggregated by user and time. The report and recommendation modules can generate statistics and recommendations without splicing multiple data sources, ensuring strong learning continuity and facilitating continuous advancement for users.

[0076] (3) The multi-service provider review and the matching of special prompts / key points for the essay improve the reliability and relevance of the review, reduce the impact of single point failure on users, and the scoring is more in line with the essay marking habits.

[0077] (4) AI writing and reviewing work together in a unified context of the essay writing. Polishing, continuation, material matching, and full-text scoring are all designed around the requirements of the essay writing, and the rewriting and suggestions are more targeted.

[0078] (5) The content generation and multi-terminal publishing of the management terminal are integrated. The content is stored in the database with a unified identifier. The user terminal and the management terminal share the same data source. Once produced, it can be displayed or pushed on multiple terminals, avoiding statistical or display deviations caused by inconsistencies in content across multiple terminals, reducing operating costs, and ensuring content consistency and timeliness.

[0079] (6) Unified data and learning reports across multiple terminals: behavioral data is associated with a unified user identifier and timestamp and written to the same storage. The user’s behavior on the mini-program and the Web forms a unified profile, supporting cross-terminal personalized recommendations and ability visualization, avoiding statistical bias caused by inconsistent data across multiple terminals, and making it convenient for users to view their learning progress anytime and anywhere.

[0080] The following describes the atomized ability training device for civil service exam preparation provided by the present invention. The atomized ability training device for civil service exam preparation described below can be referred to in correspondence with the atomized ability training method for civil service exam preparation described above.

[0081] Figure 7 This is a schematic diagram of the atomized ability training device for civil service exam preparation provided by the present invention, specifically including: The acquisition module 701 is used to acquire the user's ability training request in the civil service exam preparation system. The ability training request includes a target ability identifier and a target level identifier. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0082] The acquisition module 701 is further configured to query a preset capability training configuration table based on the target capability identifier and the target level identifier, and obtain at least one atomic question type identifier corresponding to the target capability identifier and the target level identifier, wherein the capability training configuration table is used to store the mapping relationship between capability identifiers, level identifiers and atomic question type identifiers. For detailed explanations, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0083] The extraction module 702 is used to extract corresponding atomic training questions from the atomic question bank according to the at least one atomic question type identifier. The atomic question bank stores independent training questions corresponding to each atomic question type identifier, and the independent training questions are separate from the complete real exam questions. For detailed explanation, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0084] The sending and receiving module 703 is used to send the atomic training questions to the user terminal and receive the answer data returned by the user terminal. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0085] Storage module 704 is used to associate and store the answer data with the target ability identifier, target level identifier, and corresponding atomic question type identifier in the ability training record table to obtain atomic training records. For detailed explanations, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0086] The generation module 705 is used to generate a learning report based on the atomic training records in the capability training record table. The learning report includes the completion rate and accuracy rate for each capability dimension. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0087] The generation module 705 is further configured to identify capability weaknesses based on the learning report, generate atomic training recommendation information based on the capability weaknesses, and perform atomic capability training on the user based on the atomic training recommendation information. For detailed explanations, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0088] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8 As shown, the electronic device may include: a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute an atomic ability training method for civil service exam preparation. This method includes: obtaining a user's ability training request in the civil service exam preparation system, the ability training request including a target ability identifier and a target level identifier; querying a preset ability training configuration table based on the target ability identifier and the target level identifier to obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier, wherein the ability training configuration table is used to store the mapping relationship between ability identifiers, level identifiers, and atomic question type identifiers; and extracting corresponding atomic training questions from an atomic question bank based on the at least one atomic question type identifier. The system stores independent training questions corresponding to each atomic question type identifier, and these independent training questions are separate from the complete real exam questions. The atomic training questions are sent to the user terminal, and the system receives the answer data returned by the user terminal. The answer data is associated with the target ability identifier, target level identifier, and corresponding atomic question type identifier and stored in an ability training record table to obtain atomic training records. Based on the atomic training records in the ability training record table, a learning report is generated, which includes the completion rate and accuracy rate for each ability dimension. Based on the learning report, ability weaknesses are identified, and atomic training recommendation information is generated based on these weaknesses. Atomized ability training is then performed on the user based on the atomic training recommendation information.

[0089] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0090] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the atomic ability training method for civil service exam preparation provided by the above methods. The method includes: obtaining an ability training request from a user in a civil service exam preparation system, the ability training request including a target ability identifier and a target level identifier; querying a preset ability training configuration table according to the target ability identifier and the target level identifier to obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier, wherein the ability training configuration table is used to store the mapping relationship between ability identifiers, level identifiers and atomic question type identifiers; according to the at least one atomic question type identifier... Sub-question type identifiers are used to extract corresponding atomic training questions from an atomic question bank. This atomic question bank stores independent training questions corresponding to each atomic question type identifier, and these independent training questions are separate from the complete real exam questions. The atomic training questions are sent to the user terminal, and the user terminal's response data is received. The response data is associated with the target ability identifier, target level identifier, and corresponding atomic question type identifier and stored in an ability training record table to obtain atomic training records. Based on the atomic training records in the ability training record table, a learning report is generated, which includes the completion rate and accuracy rate for each ability dimension. Based on the learning report, ability weaknesses are identified, and atomic training recommendation information is generated based on these weaknesses. Atomized ability training is then performed on the user based on the atomic training recommendation information.

[0091] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the atomic ability training method for civil service exam preparation provided by the methods described above. This method includes: obtaining an ability training request from a user in a civil service exam preparation system, the ability training request including a target ability identifier and a target level identifier; querying a preset ability training configuration table based on the target ability identifier and the target level identifier to obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier, wherein the ability training configuration table is used to store the mapping relationship between ability identifiers, level identifiers, and atomic question type identifiers; and extracting questions from an atomic question bank based on the at least one atomic question type identifier. The corresponding atomic training questions are stored in the atomic question bank, which contains independent training questions corresponding to each atomic question type identifier. These independent training questions are separate from the complete real exam questions. The atomic training questions are sent to the user terminal, and the user terminal returns the answer data. The answer data is associated with the target ability identifier, target level identifier, and corresponding atomic question type identifier and stored in the ability training record table to obtain atomic training records. Based on the atomic training records in the ability training record table, a learning report is generated, which includes the completion rate and accuracy rate under each ability dimension. Based on the learning report, ability weaknesses are identified, and atomic training recommendation information is generated based on the ability weaknesses. Atomized ability training is performed on the user based on the atomic training recommendation information.

[0092] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0093] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for training atomic ability for public examination preparation, characterized in that, include: Obtain the user's ability training request in the civil service exam preparation system, wherein the ability training request includes a target ability identifier and a target level identifier; Based on the target ability identifier and the target level identifier, a preset ability training configuration table is queried to obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier. The ability training configuration table is used to store the mapping relationship between ability identifier, level identifier and atomic question type identifier. Based on the at least one atomic question type identifier, corresponding atomic training questions are extracted from the atomic question bank, wherein the atomic question bank stores independent training questions corresponding to each atomic question type identifier, and the independent training questions are separate from the complete real questions; The atomic training questions are sent to the user terminal, and the answer data returned by the user terminal is received. The answer data is associated with the target ability identifier, the target level identifier and the corresponding atomic question type identifier and stored in the ability training record table to obtain the atomic training record; Based on the atomic training records in the capability training record table, a learning report is generated, which includes the completion rate and accuracy rate for each capability dimension. Based on the learning report, weaknesses in ability are identified, and atomic training recommendation information is generated based on these weaknesses. Atomized ability training is then performed on the user based on the atomic training recommendation information.

2. The method of claim 1, wherein, The target competency identifiers include essay writing ability identifiers or interview workshop identifiers; When the target ability identifier is the essay writing ability identifier, the atomic question type identifier includes at least one of the following: text selection, fill-in-the-blank, logical sorting, key point classification, argument splicing, title argument, essay framework, oral to written language conversion, and text compression. When the target capability identifier is an interview workshop identifier, the method further includes: Based on the interview workshop identifier and the target level identifier, query the module configuration information to obtain at least one atomic module identifier corresponding to the target level identifier. The atomic module identifier corresponds to an independent training mode and question set.

3. The method of claim 2, wherein, The step of associating and storing the answer data with the target ability identifier, target level identifier, and corresponding atomic question type identifier in the ability training record table includes: The answer data is written into the ability training record table, which includes user identifier, question identifier, ability identifier, level identifier, atomic question type identifier, score or correct / incorrect identifier, and completion timestamp. The ability identifier is used to distinguish between the essay writing ability and the interview workshop, and the atomic question type identifier or atomic module identifier is used to distinguish the specific training format.

4. The method of claim 3, wherein, The step of generating a learning report based on the atomic training records in the capability training record table includes: Receive a learning report generation request, the learning report generation request including user identifier and time range parameters; Filter the atomic training records that match the user identifier and time range parameters from the capability training record table; The system filters practice test records that match the user identifier and time range parameters from the practice test record table. The records in the practice test record table are answer data, test identifier, and completion timestamp that are written in real time in response to the user's historical practice test behavior. The practice test records and the atomic training records are stored in the same data storage. Based on the atomic training records and the set practice records, the accuracy rate under each ability dimension is calculated; Based on the accuracy rate under each capability dimension, a learning report is generated, which records radar chart data for each capability dimension.

5. The method of claim 4, wherein, The step of identifying capability weaknesses based on the learning report and generating atomic training recommendation information based on the capability weaknesses includes: Compare the accuracy rate for each capability dimension with preset thresholds; When the accuracy rate in a certain capability dimension is lower than the preset threshold, that capability dimension is determined to be a weak capability dimension. Based on the weak ability dimensions and their corresponding level identifiers, query the ability training configuration table to obtain recommended atomic question type identifiers; Based on the aforementioned weakness ability dimensions, level labels, and recommended atomic question type identifiers, atomic training recommendation information is generated.

6. The method of claim 1, wherein, The method further includes: Receive user-submitted test questions and answers; Write the answer data of the set of questions, the corresponding set of questions identifier, and the completion timestamp into the set of questions practice record table; The set of question answering data is sent to the review and scheduling module. The review and scheduling module maintains at least two AI review service instances, and each service instance is configured with a priority order. The review scheduling module calls AI review service instances sequentially according to the priority order. When a higher priority service instance is unavailable or returns an invalid result, it automatically switches to the next higher priority AI review service instance. Receive the review results, which include the total score, dimension score, key point matching list and comment information; The evaluation results are stored in the evaluation result table, and the evaluation results are associated with the corresponding records in the practice test record table.

7. The method of claim 6, wherein, The review scheduling module calls AI review service instances sequentially according to the priority order, including: Invoke the AI ​​review service instance with the current priority and send structured prompts, which include the question stem, material summary, reference points, scoring criteria and score allocation information; Receive response data and determine whether the response data meets the valid conditions. The valid conditions include that the response has not timed out, the response body can be parsed, the total score and the scores of each dimension are within the range of the question scores, and there are no missing required fields. When the response data does not meet the valid conditions, the current AI review service instance is marked as unavailable, and a call to the next priority AI review service instance is triggered. When all AI review service instances are unavailable, a default downgrade result or error message will be returned.

8. A device for training atomic ability for public examination preparation, characterized in that, include: The acquisition module is used to acquire the user's ability training request in the civil service exam preparation system. The ability training request includes a target ability identifier and a target level identifier. The acquisition module is further configured to query a preset ability training configuration table based on the target ability identifier and the target level identifier, and obtain at least one atomic question type identifier corresponding to the target ability identifier and the target level identifier, wherein the ability training configuration table is used to store the mapping relationship between ability identifier, level identifier and atomic question type identifier; An extraction module is used to extract corresponding atomic training questions from an atomic question bank based on the at least one atomic question type identifier. The atomic question bank stores independent training questions corresponding to each atomic question type identifier, and the independent training questions are separate from the complete real questions. The sending and receiving module is used to send the atomic training questions to the user terminal and receive the answer data returned by the user terminal. The storage module is used to associate the answer data with the target ability identifier, the target level identifier and the corresponding atomic question type identifier and store them in the ability training record table to obtain the atomic training record; The generation module is used to generate a learning report based on the atomic training records in the capability training record table. The learning report includes the completion rate and accuracy rate under each capability dimension. The generation module is further configured to identify capability weaknesses based on the learning report, generate atomic training recommendation information based on the capability weaknesses, and perform atomic capability training on the user based on the atomic training recommendation information.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the atomic ability training method for civil service exam preparation as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the atomic ability training method for civil service exam preparation as described in any one of claims 1 to 7.

Images