Intelligent test paper analysis and question bank automatic generation system and method

Abstract

The invention discloses an intelligent test paper analysis and question bank automatic generation system and method, and the system comprises a question recognition module which is used for analyzing question information from obtained test paper information through a large language model; the question classification module is used for classifying the question information analyzed by the question identification module; and the database management module is used for carrying out structured storage on the classified question information and outputting the corresponding question information according to the received query information. Based on the scheme disclosed by the invention, by combining advanced technologies such as image processing, natural language processing and machine learning, the question bank meeting specific requirements is automatically generated while the test paper is efficiently analyzed.

Description

Technical Field

[0001] This application relates to the field of educational technology, and particularly to an intelligent test paper analysis and question bank automatic generation system and method. Background Art

[0002] With the development of information technology, the education industry has become increasingly dependent on digital resources and automated tools. Online examinations and electronic question banks have become important components of educational assessment. However, the traditional test paper analysis and question bank generation processes mostly rely on manual operations, which are not only time-consuming and laborious but also error-prone. To improve efficiency and accuracy, a technical solution that can automatically process test papers and generate question banks is needed.

[0003] Existing technical solutions often fail to accurately identify and analyze the complex structures and formats in test papers. In addition, these tools usually cannot handle graphics, tables, and other non-text elements, which are becoming increasingly common in modern test papers.

[0004] Therefore, a more advanced technology is needed that can comprehensively analyze the content of test papers and automatically generate structured question bank data. Summary of the Invention

[0005] The purpose of this application is to provide an intelligent test paper analysis and question bank automatic generation system and method, which can, by combining advanced technologies such as image processing, natural language processing, and machine learning, efficiently analyze test papers while automatically generating a question bank that meets specific requirements.

[0006] To achieve the above purpose, this application provides the following solutions:

[0007] In a first aspect, this application provides an intelligent test paper analysis and question bank automatic generation system, which includes:

[0008] A question recognition module for parsing question information from the obtained test paper information through a large language model;

[0009] A question classification module for classifying the question information parsed by the question recognition module;

[0010] A database management module for structurally storing the classified question information and outputting corresponding question information according to the received query information.

[0011] Optionally, the question recognition module includes:

[0012] A test paper receiving unit for receiving test paper information, where the test paper information includes at least one of an electronic test paper and a scanned copy of a paper test paper;

[0013] An image processing unit for extracting text information from a scanned paper test paper through optical character recognition;

[0014] A question extraction unit for parsing the text information of the electronic test paper and / or the text information extracted from the scanned paper test paper through a large language model to identify and extract question information.

[0015] Optionally, the question extraction unit is further configured to identify problems existing in the question information through the large language model and output modification suggestions according to the problems.

[0016] Optionally, the question classification module includes:

[0017] A marking unit for obtaining feature information from the question information and marking it;

[0018] A classification unit for classifying the question information according to the feature information through a pre-trained classification model.

[0019] Optionally, the feature information includes at least one of the following: knowledge points, question types, keywords, sentence structures, lengths, complexities.

[0020] Optionally, the question classification module further includes:

[0021] A classification model training unit for training the classification model according to the historical question information with marked feature information and already classified.

[0022] Optionally, the classification categories include at least one of the following: single-choice questions, multiple-choice questions, fill-in-the-blank questions, short-answer questions, essay questions, calculation questions, proof questions.

[0023] Optionally, the database management module includes:

[0024] A storage unit for structurally storing the classified question information;

[0025] A query unit for querying the corresponding question information from the storage of the storage unit according to the received query information and outputting it.

[0026] Optionally, the system further includes: a user interface module for inputting query information and viewing the output question information.

[0027] In a second aspect, the present application provides an intelligent test paper parsing and question bank automatic generation method, and the method includes:

[0028] Parsing question information from the input test paper information through a large language model;

[0029] Classifying the question information;

[0030] Structurally store the classified question information and output the corresponding question information according to the received query information.

[0031] According to the specific embodiments provided by this application, the following technical effects are disclosed by this application:

[0032] This application provides an intelligent test paper analysis and question automatic generation system and method. By combining advanced technologies such as image processing, natural language processing, and machine learning, it realizes the efficient analysis of test papers while automatically generating a question bank that meets specific requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0034] Figure 1 It is a schematic structural diagram of an intelligent test paper analysis and question bank automatic generation system provided by an embodiment of this application;

[0035] Figure 2 It is a schematic structural diagram of a question recognition module provided by an embodiment of this application;

[0036] Figure 3 It is a schematic structural diagram of a question classification module provided by an embodiment of this application;

[0037] Figure 4 It is a schematic structural diagram of a database management module provided by an embodiment of this application; Figure 5 It is a schematic flowchart of an intelligent test paper analysis and question bank automatic production method provided by an embodiment of this application; Figure 6 It is a schematic flowchart of the process of analyzing questions in an intelligent test paper provided by an embodiment of this application; Figure 7 It is a schematic flowchart of the process of splitting questions provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] The following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only some of the embodiments of this application, rather than all of them. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of this application.

[0039] To make the above objects, features, and advantages of the present application more apparent and understandable, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0040] Refer to Figure 1 As shown, an intelligent test paper analysis and question bank automatic generation system provided by the present application includes: a question recognition module, a question classification module, and a database management module. Among them, the question classification module is respectively connected to the question recognition module and the database management module.

[0041] In an embodiment of the present application, the intelligent test paper analysis and question bank automatic generation system obtains test paper information through the question recognition module, and analyzes question information from the obtained test paper information through a large language model. Further, the parsed questions are classified by the question classification module into categories such as single-choice questions, multiple-choice questions, fill-in-the-blank questions, short-answer questions, essay questions, calculation questions, and proof questions, and the classified question information is structurally stored through the database management module, and the corresponding question information is retrieved from the stored question information according to the received query information and output.

[0042] Refer to Figure 2 As shown, the question recognition module of the present application first receives test paper information through the test paper receiving unit. Among them, the test paper information can be an editable electronic test paper such as in word format, or a test paper obtained by scanning or photographing a paper test paper.

[0043] When receiving a scanned copy or picture of a paper test paper, the image processing unit uses optical character recognition technology to process the scanned copy or picture of the paper test paper and extract the text information in the scanned copy or picture. In this embodiment, when the image processing unit extracts file information from the scanned copy or picture, it further corrects the inclination of the extracted text information to ensure the accuracy of the recognized text information; at the same time, it analyzes complex structures and formats to ensure that non-text elements such as graphics, tables, superscripts, subscripts, underlines, and mathematical formulas in the scanned copy or picture can be correctly processed.

[0044] After extracting text information from scanned documents or images, the question extraction unit uses a large language model to parse the text information in the electronic test paper or the text information extracted from the scanned document or image of the paper test paper, and identifies and extracts the question information therefrom. In this embodiment, the large language model parses and identifies the question information by using natural language processing techniques, such as semantic analysis and context understanding, to identify and understand the complex structure of the question from the text information, including: the question stem, options, answers, and possible additional instructions. In a scenario of one embodiment, the large language model uses natural language processing techniques to split the question text into individual words or phrases, removes common words that are not helpful for understanding the question, such as "de", "shi", "zai", etc., identifies the part of speech of each word (noun, verb, adjective, etc.), which helps to understand the grammatical role of the word in the sentence. Further, analyze the dependency relationship between the words in the sentence to determine the sentence structure. Understand the grammatical structure of the sentence and identify components such as the subject, predicate, and object. Further, identify the option part in the question, which is usually located after the question stem, in a specific format (such as A, B, C, D, etc.), and identify the correct answer by analyzing the question and the options.

[0045] In this embodiment, during the optimization of the parsing process, more diverse data sets are used to train the model, including test papers of different disciplines and different formats. Continuously train and optimize the machine learning model to improve the understanding of the question structure, type, and content. In this embodiment, it is also necessary to improve the quality of data annotation to ensure that the model can learn from high-quality annotations.

[0046] In this embodiment, during the process of parsing the text information, the large language model identifies possible errors or ambiguities in the question information by comparing the parsed result language text. Specifically, use a grammar analysis tool to identify the integrity and correctness of the sentence structure, integrate a spelling check algorithm to identify and suggest corrections for spelling mistakes, identify and verify the format of the question through preset format rules, and use logical reasoning ability to analyze the logical relationship between the question and the options. Further, modify error content such as missing or text errors.

[0047] In an embodiment of the present application, during the process of parsing and identifying question information through a large language model, further, a dynamic tuning and retry mechanism fed back by the large language model allows the system to adjust strategies in real time during the parsing process. When encountering difficult-to-parse questions or unsatisfactory parsing results, it can automatically adjust parsing parameters or strategies and re-parse. Specifically, by maintaining a parsing rule library containing parsing rules for various question types, the parsing rules in the parsing rule library are updated or adjusted in real time according to the feedback during the parsing process. When a parsing error is identified, the cause of the error is analyzed, and the parsing strategy is adjusted according to the error type. For example, if it is found that the recognition error rate of a certain question type is relatively high, the system may enhance the parsing algorithm for this type of question. Further, a series of parameters, such as divergence and data reference degree, are used during the parsing process, and these parameters can be adjusted in real time according to the accuracy of the parsing result. Further, an online learning mechanism can also be integrated to allow the large language model to continuously learn and optimize during the parsing process. When the model encounters uncertainty when parsing new questions, the model parameters can be adjusted immediately. In an embodiment of the present application, users are also allowed to provide feedback during the parsing process, such as confirming or correcting the parsing result. According to the user's feedback, the parsing strategy can be adjusted immediately to improve the accuracy of subsequent parsing. When judging whether the parsing result is unsatisfactory, it is jointly judged by the system and humans. Specifically, the system can compare the parsing result with the known correct answer or standard format to check for consistency. If inconsistency is found, the system may consider the parsing result to be unsatisfactory. The system can design a feedback loop to allow users to evaluate the parsing result. If the user gives negative feedback, the system can learn this feedback and use it as an indicator of unsatisfactory parsing.

[0048] Reference Figure 3 As shown, the question classification module provided by the embodiment of the present application obtains feature information from the question information parsed and identified by the marking unit and marks it. For example, feature information such as knowledge points, question types, keywords, sentence structures, length, and complexity is extracted from the question information, and based on these feature information, the question information is marked. Further, the extracted feature information is input into the classification model of the classification unit, and the question information is classified according to the marked feature information.

[0049] In an embodiment of the present application, the classification model is pre-trained by a classification model training unit according to the historical labeled feature information and the classified question information. Specifically, first, a large number of parsed and manually annotated question data are collected. These annotation information includes the knowledge points to which the questions belong (such as specific knowledge points in algebra, geometry, etc. in mathematics), question types (such as multiple-choice questions, fill-in-the-blank questions, short-answer questions, etc.). These annotated data serve as the training set for machine learning algorithms. Secondly, various features are extracted from the question text. These features can include lexical features (such as keywords and terms specific to a particular discipline), syntactic features (sentence structure, grammatical relationships, etc.), semantic features (semantic relationships between words, context semantics, etc.). Thirdly, the collected annotated data and the extracted features are used to train machine learning algorithms (such as decision trees, neural networks, etc.). The machine learning algorithm constructs a classification model by learning the relationship between the features and the annotated classifications. During the training process, the algorithm continuously adjusts the model parameters to minimize the error between the predicted classification and the actual annotated classification. For example, the decision tree algorithm constructs a classification tree based on the importance of the features, and the neural network algorithm optimizes the classification effect by adjusting the connection weights between neurons.

[0050] In this embodiment, when the classification model classifies question information, it can classify according to the knowledge point classification rules. Specifically, it classifies according to the subject knowledge system. For example, in mathematics, it is divided into large knowledge blocks such as algebra, geometry, functions, statistics, etc. Each block is further subdivided into specific knowledge points. For questions involving multiple knowledge points, they are classified according to the main knowledge point or the core examination point of the question. For example, a comprehensive question involves function and geometry knowledge. If the function knowledge is the key to solving the problem, it is classified into the relevant category under the function knowledge point. It can also classify according to the question type classification rules. Specifically, it classifies according to the question asking method and the answering requirements. For example: multiple-choice questions that provide several options and require selecting the correct answer, including single-choice questions and multiple-choice questions; fill-in-the-blank questions that require filling in the correct answer in the blank; short-answer questions that need to briefly answer questions, such as "Briefly describe the triangle interior angle sum theorem"; essay questions that require elaborating on viewpoints, principles, etc. in detail; calculation questions that mainly involve mathematical calculation processes, such as "Calculate".

[0051] Reference Figure 4As shown, the database management module of the present application structurally stores the classified question information through the storage unit. In this embodiment, a separate data model is specifically created for each question information. The data model structure includes a basic information part and a question content part. The basic information part includes, for example: question ID: each question has a unique identification number for accurately identifying and managing questions in the question bank; question type: clearly marks which type the question belongs to; creation time and modification time: records the time when the question was first created and the time of the last modification. The data model structure also includes the question content part such as: question stem: stores the question description in text form; options (for multiple-choice questions): if it is a multiple-choice question, each option is stored in sequence; answer part: correct answer: for objective questions (such as multiple-choice questions and fill-in-the-blank questions), the correct answer content is directly stored; knowledge point association part: knowledge point label: details the knowledge points involved in the question. Finally, it is stored in the memory in JSON format or XML format.

[0052] The database management module also receives the input query information through the query unit, and searches for the corresponding question information in the storage unit for output.

[0053] In the embodiment of the present application, the intelligent test paper analysis and question bank automatic generation system of the present application further includes a user interface module, which facilitates the user to input query information and display the question information fed back by the database management module for the user to view.

[0054] Reference Figure 5 As shown, the intelligent test paper analysis and question bank automatic generation method provided by the embodiment of the present application includes:

[0055] Parse the question information from the input test paper information through the large language model;

[0056] In the embodiment of the present application, after the user uploads the test paper information, the format of the test paper information is checked. When the uploaded test paper is in the format of a scanned paper test paper or a picture format, the optical character recognition technology (OCR) is preferably used to process the scanned copy or picture, extract the text information in the scanned copy or picture, and further perform preprocessing operations such as noise removal and skew correction on the extracted text information to ensure the recognition accuracy, and perform parsing for complex structures and formats to ensure that non-text elements such as graphics, tables, superscripts, subscripts, underlines, and mathematical formulas can also be correctly processed. Further, the question information is parsed and recognized from the preprocessed text information through the large language model. When the uploaded test paper is an electronic test paper in word format, editable PDF format, etc., the large language model directly uses natural language processing technology to perform semantic analysis and context understanding on the text in the electronic test paper, and recognizes and understands the question information from the text information.

[0057] Reference Figure 6 As shown, for the process of intelligently analyzing test questions provided by the embodiments of the present application, after the user uploads the test paper document, first, the document format of the test paper is detected and processed. When the document format of the test paper is.doc, it is converted to.docx, and the text content of the test paper is directly read from the converted.docx format document; when the document format of the test paper is in picture format or scanned copy format, the optical character recognition technology is used to read the text content of the test paper from the document in picture format or scanned copy format. Further, preprocessing such as noise removal and skew correction is performed on the read text content to ensure accurate recognition of the text content. At the same time, the complex structures and formats in the text content are parsed to ensure that non-text elements such as graphics, tables, superscripts, subscripts, underlines, and mathematical formulas in the test paper document can be correctly processed. Further, the questions in the text information of the test paper are split to distinguish the content of each question. When parsing and recognizing the content of each question, different large language models are used for different question types to parse the question into corresponding parts: such as title, options, answers, analysis, etc. Further, the analysis results are checked, and the text containing special formats such as tables, pictures, superscripts, subscripts, underlines, and mathematical formulas is marked as html format, and all the information of the same question is integrated to generate a json structure of the test paper for storage. When receiving a viewing request, the front end renders the preview page of the question according to the json structure.

[0058] Reference Figure 7 As shown, for the question splitting process provided by the embodiments of the present application, when splitting the questions in the text information of the test paper, line numbers are added to the text, the large language model is used to identify and split the main content of the test paper, and the title and line number range of each part are returned. Further, the line number range is verified. If it is found that there are missing lines or duplicate lines in the split line number range, the problems found are sorted out and passed to the large language model. The large language model re-identifies and splits the main content of the test paper by adjusting the parsing strategy until the verification is successful. After obtaining the main content of the test paper and the corresponding line numbers, each main part of the test paper is processed separately, and the large language model is used again to split, distinguishing the question and non-question content, and giving the corresponding complete text. Further verify whether the content returned by the large language model is consistent with the original text. If the verification is successful, the specific content of each question is obtained. If the verification fails, it is determined that there are differences between the content returned by the large language model and the original text. The problems existing in the content returned by the large language model are sorted out and fed back to the large language model together. The parsing strategy is adjusted, and the questions and non-question content are split again until the verification is successful.

[0059] Classify the question information;

[0060] In an embodiment of the present application, after identifying and extracting question information from the test paper information, feature information of the question, such as knowledge points, question types, keywords, sentence structures, length, and complexity, etc., is extracted from the question information, and the question information is marked based on these feature information. Further, the classification model classifies the question information through these feature information.

[0061] In this embodiment, during specific classification, classification is carried out according to the knowledge point classification rules, specifically classified according to the subject knowledge system. For example, in the mathematics subject, classification is carried out according to large knowledge sections such as algebra, geometry, functions, and statistics. Among them, specific knowledge points can be further subdivided under each section. For questions involving multiple knowledge points, classification is carried out according to the main knowledge point or the core examination point of the question. For example, a comprehensive question involves function and geometry knowledge points. If the function knowledge point is the key to solving the problem, it is classified into the relevant category under the function knowledge point; if the geometry knowledge point is the key to solving the problem, it is classified into the relevant category under the geometry knowledge point. In this embodiment, during specific classification, classification is carried out according to the question type classification rules, specifically classified according to the question's asking method and answering requirements. For example, multiple-choice questions that require selecting the correct option by providing several options, including single-choice questions and multiple-choice questions; fill-in-the-blank questions that require filling in the correct answer in the blank; short-answer questions that require briefly answering questions, such as "Briefly describe the triangle interior angle sum theorem"; essay questions that require elaborating on viewpoints, principles, etc. in detail; calculation questions mainly involving mathematical calculation processes, such as "Calculate".

[0062] The classified question information is stored in a structured manner, and the corresponding question information is output according to the received query information.

[0063] In the embodiments of the present application, after classifying the parsed and extracted question information, the question information is stored in a structured manner. In this embodiment, first, an independent data model is created for each piece of question information, including multiple fields to store information on different aspects of the question. The structure of this data model includes a basic information part and a question content part. Among them, the basic information part includes: a question ID field, each question has a unique identification number, which is used to accurately identify and manage questions in the question bank; a question type field, which is used to clearly mark which type of question it belongs to; creation time and modification time fields, which are used to record the time when the question was first created and the time of the last modification. Among them, the question content part includes: a question stem field, which stores the complete question description in text form; an option field (for multiple-choice questions), if it is a multiple-choice question, each option is stored in sequence; an answer part field, correct answer: for objective questions (such as multiple-choice questions, fill-in-the-blank questions), the correct answer content is directly stored; a knowledge point association part field, knowledge point tags, which are used to list in detail the knowledge points involved in the question. After creating the data model of the question, it is stored in JSON format or XML format to form a question bank. In the embodiments of the present application, when receiving the query information input by the user, the query information is queried and matched with the question information in the question bank for output to facilitate the user to view. In this embodiment, the user can also view the detailed analysis of each question and manually correct the misidentified parts. At the same time, the data in the question bank is edited and maintained, such as adding new questions, deleting old questions, or updating question information.

[0064] In summary, through the solution provided by the present application, the following innovations are achieved:

[0065] 1. Comprehensive test paper analysis: The present invention can analyze text, graphics, tables, and other non-text elements in Word documents, providing comprehensive test paper analysis capabilities.

[0066] 2. Automated question bank generation: Through an automated process, manual intervention is reduced, improving the efficiency and quality of question bank construction.

[0067] 3. User-friendly interaction interface: Provide an intuitive user interaction interface to enhance the user experience.

[0068] 4. Flexible data structuring: It can flexibly convert the parsed data into multiple question bank formats to meet different online exam and assessment requirements.

[0069] In the embodiments of the present application, the question information in the question bank can also be exported in multiple formats, facilitating integration with other systems or docking with different exam platforms.

[0070] In an embodiment of the present application, the question information in the question bank is backed up and the version of the question bank is controlled to ensure the security and reliability of the question bank.

[0071] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0072] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0073] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0074] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0075] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the various embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random access memories (ReRAM), magnetoresistive random access memories (MRAM), ferroelectric random access memories (FRAM), phase change memories (PCM), graphene memories, etc. Volatile memories can include random access memory (RAM) or external cache memories, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0076] The databases involved in the various embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the various embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0077] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0078] Specific examples are used in this article to elaborate on the principles and implementation manners of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. An intelligent test paper analysis and question bank automatic generation system, characterized in that: The system comprises: The question recognition module is used to parse the question information from the obtained test paper information through a large language model; A topic classification module, used for classifying the topic information parsed by the topic identification module; The database management module is used to store the classified topic information in a structured manner and output the corresponding topic information according to the received query information.

2. The intelligent test paper analysis and question bank automatic generation system according to claim 1 is characterized in that: The topic identification module comprises: The test paper receiving unit is used to receive test paper information, wherein the test paper information includes at least one of an electronic test paper and a scanned copy of a paper test paper; An image processing unit, used for extracting text information from a scanned paper test paper by optical character recognition; The question extraction unit is used to parse the text information of the electronic test paper and / or the text information extracted from the scanned copy of the paper test paper through a large language model to identify and extract question information.

3. The intelligent test paper analysis and question bank automatic generation system according to claim 2 is characterized in that: The topic extraction unit is also used to identify problems existing in the topic information through the large language model, and output modification suggestions based on the problems.

4. The intelligent test paper analysis and question bank automatic generation system according to any one of claims 1 to 3, characterized in that: The topic classification module includes: A marking unit, used for obtaining characteristic information from the topic information and marking it; A classification unit is used to classify the topic information according to the feature information using a pre-trained classification model.

5. The intelligent test paper analysis and question bank automatic generation system according to claim 4 is characterized in that: The feature information includes at least one of the following: knowledge point, question type, keyword, sentence structure, length, and complexity.

6. The intelligent test paper analysis and question bank automatic generation system according to claim 4 is characterized in that: The topic classification module also includes: The classification model training unit is used to train the classification model according to the historically marked feature information and classified topic information.

7. The intelligent test paper analysis and question bank automatic generation system according to any one of claims 1 to 3, characterized in that: The classification categories include at least one of the following: single-choice questions, multiple-choice questions, fill-in-the-blank questions, short-answer questions, essay questions, calculation questions, and proof questions.

8. The intelligent test paper analysis and question bank automatic generation system according to any one of claims 1 to 3, characterized in that: The database management module includes: A storage unit, used for structurally storing the classified topic information; The query unit is used to search for corresponding topic information from the storage unit according to the received query information and output it.

9. The intelligent test paper analysis and question bank automatic generation system according to any one of claims 1 to 3, characterized in that: The system further comprises: The user interface module is used to input query information and view output question information.

10. A method for intelligent test paper analysis and automatic question bank generation, characterized in that: The method comprises: Parse the question information from the input test paper information through a large language model; Classifying the topic information; The classified topic information is stored in a structured manner, and the corresponding topic information is output according to the received query information.

Images