Question drawing and test paper composition method and system based on big data and big model

Abstract

The embodiment of the present application discloses a method and system for drawing questions and composing papers based on big data and big models, belonging to the field of learning and education technology, including: establishing a general question bank and a classified question bank, storing historical real questions in the classified question bank according to the classification results, establishing a test paper bank, extracting features from historical test papers, establishing an initial test paper template based on the features extracted from the historical test papers, adjusting the initial test paper template to obtain a test paper template, inputting an examination syllabus and test paper composition requirements to obtain constraints, searching and selecting corresponding historical real questions, establishing multiple modification and adjustment models, calling the modification and adjustment models according to needs to modify and adjust the historical real questions to generate new questions, filling the selected historical real questions and the generated new questions into the test paper template to generate an examination paper. The embodiment of the present application can prevent examinees from answering questions by rote memorization, so that the test paper can accurately test the examinee's knowledge level.

Description

Technical Field

[0001] The present application belongs to the field of learning and educational technology, and specifically, to a method and system for selecting questions and assembling examination papers based on big data and big models. Background Art

[0002] Traditional exercises and test papers are often conducted through manual selection and production, which is time-consuming and lacks accurate control over the test question structure. With the development of software engineering technology and network technology, some of the drawbacks of traditional tests have been solved. Online homework, online tests, and in-class assessments have become widely used in daily learning life, making exploratory attempts in test reform and playing a positive role.

[0003] One of the prerequisites for the implementation of online examinations is the construction of question banks and the setting of questions. However, the existing question banks are not highly standardized and the paper-forming algorithms are unscientific. It is easy for repeated questions or questions with high similarity in knowledge points to appear in the same paper. The quality of the paper is often not very high. In addition, the existing question banks also obtain data from already public questions, and candidates can also search for these public questions. For example, if the knowledge points tested in the paper are relatively niche, there are fewer corresponding questions in the question bank. Candidates only need to practice a large number of corresponding questions, and they are likely to encounter questions in the question bank. They only need to memorize them without understanding and digesting the knowledge points to get high scores. This runs counter to the original purpose of the exam and fails to grasp and understand the candidate's original knowledge level. Summary of the invention

[0004] In order to solve the above problems and technical defects, the embodiment of the present application adopts the following technical solution, which is a method for drawing questions and composing examination papers based on big data and big models, including the following steps:

[0005] Step 1: Establish a general question bank, obtain historical real questions and parameter attributes of historical real questions, classify the historical real questions into knowledge points according to the parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical real questions in the classified question bank according to the classification results;

[0006] Step 2: Establish a test paper database, obtain historical test papers, extract features from historical test papers, obtain test paper types, question types in the test papers, the number of questions and score proportions of each question type, and establish an initial test paper template based on the features extracted from historical test papers;

[0007] Step 3: The user calls the initial test paper template, adjusts the initial test paper template according to the requirements, obtains the test paper template, inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements;

[0008] Step 4: Search the classified question bank corresponding to the knowledge points in the general question bank according to the examination syllabus, and search and select the corresponding historical questions in the classified question bank according to the constraints;

[0009] Step 5: Establish multiple modification and adjustment models, and call one or more modification and adjustment models to modify and adjust historical questions to generate new questions according to needs;

[0010] Step 6: Fill the selected historical questions and generated new questions into the test paper template according to the constraints to generate the test paper.

[0011] Preferably, the parameter attributes of the historical real questions include: knowledge test points, question types, difficulty coefficient, number of historical uses and classic excellence;

[0012] Question types include: single-choice questions, multiple-choice questions, true-or-false questions, fill-in-the-blank questions, calculation and solution questions, reading comprehension questions, essay questions and listening questions;

[0013] The difficulty coefficient is obtained based on the average score rate of each historical real question.

[0014] The number of historical usages is when the target historical questions are stored in the total question bank. The initial number of historical usages is one. Every time the target historical questions are called, the number of historical usages increases by one.

[0015] The classic excellence degree refers to the coordination of the test design of the target history real test questions, the novelty of the questions, the test coverage of the knowledge test points and the degree of deviation from the knowledge test points. It is rated by the user after use. The ratings include: excellent, good and mediocre. When multiple ratings are included, the mode of the ratings is taken as the classic excellence degree of the target history real test questions.

[0016] Furthermore, the steps of classifying knowledge points of historical questions according to parameter attributes are as follows:

[0017] Obtain the text of the questions, the answer process and the analysis of the questions, and perform the first classification and word segmentation processing on the text of the questions, the answer process and the analysis of the questions;

[0018] Construct a feature word model based on the first classification result, and calculate the feature word weight based on the word segmentation result;

[0019] Calculate the similarity between each historical question based on the feature word model and feature word weight;

[0020] According to the similarity calculation results, the historical real questions are classified for the second time to obtain the knowledge test point classification results.

[0021] Preferably, the examination outline is all the knowledge points of the entire examination paper; the constraints include: examination paper difficulty coefficient, examination outline, number of each question type, proportion of each knowledge point in each question type, total examination paper score, and score proportion of each question type.

[0022] Furthermore, the search and selection of corresponding historical questions in the classified question bank according to the constraint conditions is to search for corresponding historical questions according to the difficulty coefficient. The user enters the difficulty coefficient of each knowledge point in the test paper and the proportion of each knowledge point in each question type in the test paper composition requirements to obtain the number of questions and the difficulty coefficient of the questions in the current classified question bank. The difficulty coefficient calculation formula is as follows:

[0023] P = (∑Mi × Ki) / ∑Ki;

[0024] Where i=1, 2, ...N, N is the number of questions within the question type range, P is the difficulty coefficient of the input question type, Mi is the difficulty coefficient of the i-th question, and Ki is the score of the i-th question.

[0025] Furthermore, when there are multiple real history questions that meet the constraint conditions, they are compared based on the rating of classic excellence. Real history questions rated as excellent have higher priority than real history questions rated as good and mediocre, and real history questions rated as good have higher priority than real history questions rated as mediocre.

[0026] When the ratings of classic excellence are equal, a comparison is made based on the number of historical uses, and the historical question with the smallest number of historical uses is selected.

[0027] Further, the modification and adjustment model includes: a selection order replacement model, a conditional answer exchange model, a numerical value reset model and a multi-question reorganization model;

[0028] The selection order replacement model is used to replace the order of options in multiple-choice questions and generate new questions and new answers after the selection order is replaced;

[0029] The conditional answer exchange model is used to exchange the given condition value and the answer value in the question. The answer value is given in the question, and the original answer value is used as the new condition value. The original condition value is used as the new answer value, and the original condition value is solved to vertically generate a new question and a new answer according to the new condition.

[0030] The numerical reset model is used to obtain the calculation formula in the original historical real questions, randomly reset the original conditional values ​​when the requirements of the questions are met, input the reset conditional values ​​into the calculation formula, obtain the new answer results, and generate new questions and new answers based on the new conditional values;

[0031] The multi-question recombination model is used to combine multiple historical questions within the same knowledge test point. First, the calculation formula in the original historical question is obtained, and multiple historical questions that call the same calculation formula are obtained. The causal relationship between the conditional values ​​and answer values ​​in each historical question and the conditional values ​​and answer values ​​in other historical questions is determined, and the conditional values ​​and answer values ​​with the same meaning in multiple historical questions are modified and unified. The modified conditional values ​​and answer values ​​are then input into the calculation formula for calculation verification, and new questions and new answers are generated based on the verified conditional values ​​and answer values.

[0032] The question-drawing and test-taking system based on big data and big models includes:

[0033] The question bank module is used to obtain historical real questions and parameter attributes of historical real questions, classify the historical real questions into knowledge test points according to the parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical real questions in the classified question bank according to the classification results;

[0034] The paper library module is used to establish a paper library, obtain historical papers, extract features from historical papers, obtain the paper type, the types of questions in the paper, the number of questions and the score of each type of question, and establish the initial paper template based on the features extracted from the historical papers;

[0035] The template condition module is used for the user to call the initial template of the test paper, adjust the initial template of the test paper according to the needs, and obtain the test paper template. The user inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements;

[0036] The question acquisition module is used to search the classified question bank corresponding to the knowledge test points in the general question bank according to the examination syllabus, search and select the corresponding historical questions in the classified question bank according to the constraint conditions, establish multiple modification and adjustment models, and call one or more modification and adjustment models according to the needs to modify and adjust the historical questions to generate new questions;

[0037] The test paper generation module is used to fill the selected historical real questions and generated new questions into the test paper template according to the constraints to generate the test paper.

[0038] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the method for selecting questions and assembling examination papers based on big data and big models as described above are implemented.

[0039] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for selecting questions and composing examination papers based on big data and big models as described above.

[0040] Compared with the prior art, the beneficial effects of the embodiments of the present application are:

[0041] The present application establishes a classified question bank by classifying knowledge test points in the general question bank, and stores historical real questions in the classified question bank according to the classification results, which makes it convenient, fast and accurate to screen and extract questions later. At the same time, multiple modification and adjustment models are set up to adjust and modify the questions in the existing classified question bank when extracting questions. Without changing the test points of the original questions, the modified questions are very different from the original historical real questions, but have the same knowledge test points. Only when the examinees fully understand the knowledge test points can they answer the modified questions. While drawing questions and composing papers, it also avoids the situation where examinees answer questions by rote memorization, so that the test paper can accurately test the examinees' knowledge level. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In the attached picture:

[0043] Figure 1 A schematic diagram of the method steps of an embodiment of the present application;

[0044] Figure 2 Schematic diagram of the system structure of an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Generally, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0046] Example 1

[0047] like Figure 1 As shown in the figure, the method of drawing questions and setting up examination papers based on big data and big models includes the following contents:

[0048] Establish a general question bank, obtain historical real questions and parameter attributes of historical real questions, classify the historical real questions into knowledge points according to the parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical real questions in the classified question bank according to the classification results;

[0049] The parameter attributes of historical real questions include: knowledge points, question types, difficulty coefficient, number of historical uses and classic excellence;

[0050] Question types include: single-choice questions, multiple-choice questions, true-or-false questions, fill-in-the-blank questions, calculation and solution questions, reading comprehension questions, essay questions and listening questions;

[0051] The difficulty coefficient is obtained based on the average score rate of each historical real question.

[0052] The number of historical usages is when the target historical questions are stored in the total question bank. The initial number of historical usages is one. Every time the target historical questions are called, the number of historical usages increases by one.

[0053] The classic excellence degree refers to the coordination of the test design of the target history real test questions, the novelty of the questions, the test coverage of the knowledge test points and the degree of deviation from the knowledge test points. It is rated by the user after use. The ratings include: excellent, good and mediocre. When multiple ratings are included, the mode of the ratings is taken as the classic excellence degree of the target history real test questions.

[0054] The steps of classifying knowledge points of historical questions according to parameter attributes are as follows:

[0055] Obtain the text of the questions, the answer process and the analysis of the questions, and perform the first classification and word segmentation processing on the text of the questions, the answer process and the analysis of the questions;

[0056] Construct a feature word model based on the first classification result, and calculate the feature word weight based on the word segmentation result;

[0057] Calculate the similarity between each historical question based on the feature word model and feature word weight;

[0058] According to the similarity calculation results, the historical real questions are classified for the second time to obtain the knowledge test point classification results.

[0059] Establish a test paper database, obtain historical test papers, extract features from historical test papers, obtain test paper types, question types in the test papers, the number of questions and score ratios of each question type, and establish an initial test paper template based on the features extracted from historical test papers.

[0060] The user calls the initial test paper template, adjusts the initial test paper template according to the requirements, and obtains the test paper template. The user inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements;

[0061] The examination syllabus includes all the knowledge points of the entire examination paper; the constraints include: the difficulty coefficient of the examination paper, the examination syllabus, the number of each question type, the proportion of each knowledge point in each question type, the total score of the examination paper, and the proportion of each question type.

[0062] Search the classified question bank corresponding to the knowledge points in the general question bank according to the examination syllabus, and search and select the corresponding historical questions in the classified question bank according to the constraints;

[0063] The searching and selecting of corresponding historical questions in the classified question bank according to the constraint conditions is to search for corresponding historical questions according to the difficulty coefficient. The user inputs the difficulty coefficient of each knowledge point in the test paper and the proportion of each knowledge point in each question type in the test paper composition requirements, and obtains the number of questions and the difficulty coefficient of the questions in the current classified question bank. The difficulty coefficient calculation formula is as follows:

[0064] P = (∑Mi×Ki) / ∑Ki;

[0065] Where i=1, 2, ...N, N is the number of questions within the question type range, P is the difficulty coefficient of the input question type, Mi is the difficulty coefficient of the i-th question, and Ki is the score of the i-th question.

[0066] When there are multiple real history questions that meet the constraints, they are compared based on the classic excellence rating. Real history questions rated as excellent have higher priority than real history questions rated as good and mediocre, and real history questions rated as good have higher priority than real history questions rated as mediocre.

[0067] When the ratings of classic excellence are equal, a comparison is made based on the number of historical uses, and the historical question with the smallest number of historical uses is selected.

[0068] Establish multiple modification and adjustment models, and call one or more modification and adjustment models to modify and adjust historical questions to generate new questions according to needs;

[0069] The modification and adjustment models include: a selection order replacement model, a conditional answer exchange model, a numerical value reset model and a multi-question reorganization model;

[0070] The selection order replacement model is used to replace the order of options in multiple-choice questions and generate new questions and new answers after the selection order is replaced;

[0071] The conditional answer exchange model is used to exchange the given condition value and the answer value in the question. The answer value is given in the question, and the original answer value is used as the new condition value. The original condition value is used as the new answer value, and the original condition value is solved to vertically generate a new question and a new answer according to the new condition.

[0072] The numerical reset model is used to obtain the calculation formula in the original historical real questions, randomly reset the original conditional values ​​when the requirements of the questions are met, input the reset conditional values ​​into the calculation formula, obtain the new answer results, and generate new questions and new answers based on the new conditional values;

[0073] For example, in a physics problem, "the object falls on a planet for 10 seconds, and the gravitational acceleration g is 10 m / s 2 , calculate the distance h" that the object falls from a height;

[0074] The formula for free fall motion is:

[0075] h=gt 2 / 2

[0076] The answer is h = 500m;

[0077] After the condition values ​​are randomly reset, t = 5S, g = 30m / s 2 , the answer is h = 375m;

[0078] The new question is "The object falls on a planet for 5 seconds, and the gravitational acceleration g is 30m / s 2 , calculate the distance h" that the object falls from a height.

[0079] The multi-question recombination model is used to combine multiple historical questions within the same knowledge test point. First, the calculation formula in the original historical question is obtained, and multiple historical questions that call the same calculation formula are obtained. The causal relationship between the conditional values ​​and answer values ​​in each historical question and the conditional values ​​and answer values ​​in other historical questions is determined, and the conditional values ​​and answer values ​​with the same meaning in multiple historical questions are modified and unified. The modified conditional values ​​and answer values ​​are then input into the calculation formula for calculation verification, and new questions and new answers are generated based on the verified conditional values ​​and answer values.

[0080] Similarly, in a physics question, "The radius of a planet is r, its mass is M, and its gravitational constant is G. Find the gravitational acceleration g of the planet";

[0081] The law of universal gravitation is used:

[0082] g=GM / r 2

[0083] The above question is "The object falls on a planet for 10 seconds, and the gravitational acceleration g is 10m / s 2 , calculate the distance h that an object falls from a height". The conditional value g in the above question has the same meaning as the answer value g in the second question. Adjust the two verticals to unify the two values ​​g. At the same time, modify other values ​​according to the unified value g, and get "the radius of a planet is r, the mass is M, the gravitational constant is G, and the time an object falls on the planet is t. Find the gravitational acceleration g of the planet and the distance h that an object falls from a height".

[0084] After using the multi-question reorganization model to reorganize the questions, you can also call the conditional answer exchange model and the numerical reset model. For example, use the conditional answer exchange model to exchange the distance h and time t in the reorganized question, and the question becomes "Find the gravitational acceleration g of the planet and the time t when an object falls from a height." Use the numerical reset model to modify the conditional values ​​in "a planet with a radius of r, a mass of M, a gravitational constant of G, and an object falling distance h from the planet" to obtain new answer values ​​"gravitational acceleration g" and "time t." Finally, generate a new question based on the original question.

[0085] According to the constraints, the selected historical questions and the generated new questions are filled into the test paper template to generate the test paper.

[0086] Example 2

[0087] like Figure 2 As shown in the figure, the question drawing and test paper composition system based on big data and big models includes:

[0088] The question bank module is used to obtain historical real questions and parameter attributes of historical real questions, classify the historical real questions into knowledge test points according to the parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical real questions in the classified question bank according to the classification results;

[0089] The paper library module is used to establish a paper library, obtain historical papers, extract features from historical papers, obtain the paper type, the types of questions in the paper, the number of questions and the score of each type of question, and establish the initial paper template based on the features extracted from the historical papers;

[0090] The template condition module is used for the user to call the initial template of the test paper, adjust the initial template of the test paper according to the needs, and obtain the test paper template. The user inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements;

[0091] The question acquisition module is used to search the classified question bank corresponding to the knowledge test points in the general question bank according to the examination syllabus, search and select the corresponding historical questions in the classified question bank according to the constraint conditions, establish multiple modification and adjustment models, and call one or more modification and adjustment models according to the needs to modify and adjust the historical questions to generate new questions;

[0092] The test paper generation module is used to fill the selected historical real questions and generated new questions into the test paper template according to the constraints to generate the test paper.

[0093] Example 3

[0094] From the hardware level, the present application provides an embodiment of an electronic device that implements all or part of the content of the method for drawing questions and assembling examination papers based on big data and big models, and the electronic device specifically includes the following content:

[0095] Step 1: Establish a general question bank, obtain historical real questions and parameter attributes of historical real questions, classify the historical real questions into knowledge points according to the parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical real questions in the classified question bank according to the classification results;

[0096] Step 2: Establish a test paper database, obtain historical test papers, extract features from historical test papers, obtain test paper types, question types in the test papers, the number of questions and score proportions of each question type, and establish an initial test paper template based on the features extracted from historical test papers;

[0097] Step 3: The user calls the initial test paper template, adjusts the initial test paper template according to the requirements, obtains the test paper template, inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements;

[0098] Step 4: Search the classified question bank corresponding to the knowledge points in the general question bank according to the examination syllabus, and search and select the corresponding historical questions in the classified question bank according to the constraints;

[0099] Step 5: Establish multiple modification and adjustment models, and call one or more modification and adjustment models to modify and adjust historical questions to generate new questions according to needs;

[0100] Step 6: Fill the selected historical questions and generated new questions into the test paper template according to the constraints to generate the test paper.

[0101] Example 4

[0102] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the method for drawing questions and assembling examination papers based on big data and big models in the above-mentioned embodiments, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, all the steps of the method for drawing questions and assembling examination papers based on big data and big models in the above-mentioned embodiments are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0103] Step 1: Establish a general question bank, obtain historical real questions and parameter attributes of historical real questions, classify the historical real questions into knowledge points according to the parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical real questions in the classified question bank according to the classification results;

[0104] Step 2: Establish a test paper database, obtain historical test papers, extract features from historical test papers, obtain test paper types, question types in the test papers, the number of questions and score proportions of each question type, and establish an initial test paper template based on the features extracted from historical test papers;

[0105] Step 3: The user calls the initial test paper template, adjusts the initial test paper template according to the requirements, obtains the test paper template, inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements;

[0106] Step 4: Search the classified question bank corresponding to the knowledge points in the general question bank according to the examination syllabus, and search and select the corresponding historical questions in the classified question bank according to the constraints;

[0107] Step 5: Establish multiple modification and adjustment models, and call one or more modification and adjustment models to modify and adjust historical questions to generate new questions according to needs;

[0108] Step 6: Fill the selected historical questions and generated new questions into the test paper template according to the constraints to generate the test paper.

[0109] The above-mentioned embodiments only express the preferred implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for ordinary technicians in this field, several modifications, improvements and substitutions can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. The method of selecting questions and setting examination papers based on big data and big models is characterized by: The following steps are involved: Step 1: Establish a general question bank, obtain historical real questions and their parameter attributes, including knowledge points, question types, difficulty coefficient, historical usage times and classic excellence; Question types include: single-choice questions, multiple-choice questions, true-or-false questions, fill-in-the-blank questions, calculation and solution questions, reading comprehension questions, essay questions and listening questions; The difficulty coefficient is obtained based on the average score rate of each historical real question. The number of historical usages is when the target historical questions are stored in the total question bank. The initial number of historical usages is one. Every time the target historical questions are called, the number of historical usages increases by one. Classic excellence refers to the coordination of the test design of the target history test paper, the novelty of the questions, the coverage of the knowledge test points and the degree of deviation from the knowledge test points. The users will rate it after use. The ratings include: excellent, good and mediocre. When there are multiple ratings, the mode of the ratings is taken as the classic excellence of the target history test paper. Classify the knowledge points of historical questions according to parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical questions in the classified question bank according to the classification results; Step 2: Establish a test paper database, obtain historical test papers, extract features from historical test papers, obtain test paper types, question types in the test papers, the number of questions and score proportions of each question type, and establish an initial test paper template based on the features extracted from historical test papers; Step 3: The user calls the initial test paper template, adjusts the initial test paper template according to the requirements, obtains the test paper template, inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements; Step 4: Search the classification question bank corresponding to the knowledge points in the general question bank according to the examination syllabus; According to the constraints, the corresponding historical questions are searched and selected in the classified question bank, and the corresponding historical questions are searched according to the difficulty coefficient. The user enters the difficulty coefficient of each knowledge point in the test paper and the proportion of each knowledge point in each question type in the test paper composition requirements, and obtains the number of questions and the difficulty coefficient of the questions in the current classified question bank. The difficulty coefficient calculation formula is as follows: P = (∑Mi × Ki) / ∑Ki; Where i=1, 2, ...N, N is the number of questions in the question type range, P is the difficulty coefficient of the input question type, Mi is the difficulty coefficient of the i-th question, and Ki is the score of the i-th question; When there are multiple real history questions that meet the constraints, they are compared based on the classic excellence rating. Real history questions rated as excellent have higher priority than real history questions rated as good and mediocre, and real history questions rated as good have higher priority than real history questions rated as mediocre. When the ratings of classic excellence are equal, they are compared based on the number of historical uses, and the historical real questions with the smallest number of historical uses are selected; Step 5: Establish multiple modification and adjustment models, and call one or more modification and adjustment models to modify and adjust historical questions to generate new questions according to needs. The modification and adjustment models include: selection order replacement model, conditional answer replacement model, numerical reset model and multi-question reorganization model; The selection order replacement model is used to replace the order of options in multiple-choice questions and generate new questions and new answers after the selection order is replaced; The conditional answer exchange model is used to exchange the given condition value and the answer value in the question. The answer value is given in the question, and the original answer value is used as the new condition value. The original condition value is used as the new answer value, and the original condition value is solved to vertically generate a new question and a new answer according to the new condition. The numerical reset model is used to obtain the calculation formula in the original historical real questions, randomly reset the original conditional values ​​when the requirements of the questions are met, input the reset conditional values ​​into the calculation formula, obtain the new answer results, and generate new questions and new answers based on the new conditional values; The multi-question reorganization model is used to combine multiple historical questions in the same knowledge test point. First, the calculation formula in the original historical question is obtained to obtain multiple historical questions that call the same calculation formula. The causal relationship between the condition value and the answer value in each historical question and the condition value and the answer value in other historical questions is determined. The condition values ​​and answer values ​​with the same meaning in multiple historical questions are modified and unified, and then the modified condition values ​​and answer values ​​are input into the calculation formula for calculation verification, and new questions and new answers are generated according to the verified condition values ​​and answer values. Step 6: Fill the selected historical questions and generated new questions into the test paper template according to the constraints to generate the test paper.

2. The method for drawing questions and setting examination papers based on big data and big models according to claim 1 is characterized in that: The steps of classifying knowledge points of historical questions according to parameter attributes are as follows: Obtain the text of the questions, the answer process and the analysis of the questions, and perform the first classification and word segmentation processing on the text of the questions, the answer process and the analysis of the questions; Construct a feature word model based on the first classification result, and calculate the feature word weight based on the word segmentation result; Calculate the similarity between each historical question based on the feature word model and feature word weight; According to the similarity calculation results, the historical real questions are classified for the second time to obtain the knowledge test point classification results.

3. The method for drawing questions and setting test papers based on big data and big models according to claim 2 is characterized in that: The examination outline includes all the knowledge points of the entire examination paper; the constraints include: the difficulty coefficient of the examination paper, the examination outline, the number of each question type, the proportion of each knowledge point in each question type, the total score of the examination paper, and the score proportion of each question type.

4. The question-drawing and test-taking system based on big data and big models is characterized by: include: The question bank module is used to obtain historical real questions and their parameter attributes, including knowledge points, question types, difficulty coefficient, historical usage times and classic excellence; Question types include: single-choice questions, multiple-choice questions, true-or-false questions, fill-in-the-blank questions, calculation and solution questions, reading comprehension questions, essay questions and listening questions; The difficulty coefficient is obtained based on the average score rate of each historical real question. The number of historical usages is when the target historical questions are stored in the total question bank. The initial number of historical usages is one. Every time the target historical questions are called, the number of historical usages increases by one. Classic excellence refers to the coordination of the test design of the target history test paper, the novelty of the questions, the coverage of the knowledge test points and the degree of deviation from the knowledge test points. The users will rate it after use. The ratings include: excellent, good and mediocre. When there are multiple ratings, the mode of the ratings is taken as the classic excellence of the target history test paper. Classify the knowledge points of historical questions according to parameter attributes, establish a classified question bank in the general question bank according to the classification results, and store the historical questions in the classified question bank according to the classification results; The paper library module is used to establish a paper library, obtain historical papers, extract features from historical papers, obtain the paper type, the types of questions in the paper, the number of questions and the score of each type of question, and establish the initial paper template based on the features extracted from the historical papers; The template condition module is used for the user to call the initial template of the test paper, adjust the initial template of the test paper according to the needs, and obtain the test paper template. The user inputs the test outline and test paper composition requirements, and obtains the constraint conditions according to the test paper template and test paper composition requirements; The question acquisition module is used to search the classified question bank corresponding to the knowledge test point in the general question bank according to the examination syllabus, search and select the corresponding historical real questions in the classified question bank according to the constraint conditions, and search for the corresponding historical real questions according to the difficulty coefficient. The user enters the difficulty coefficient of each knowledge test point in the test paper and the proportion of each knowledge test point in each question type in the test paper composition requirements to obtain the number of questions and the difficulty coefficient of the questions in the current classified question bank. The difficulty coefficient calculation formula is as follows: P = (∑Mi × Ki) / ∑Ki; Where i=1, 2, ...N, N is the number of questions in the question type range, P is the difficulty coefficient of the input question type, Mi is the difficulty coefficient of the i-th question, and Ki is the score of the i-th question; When there are multiple real history questions that meet the constraints, they are compared based on the classic excellence rating. Real history questions rated as excellent have higher priority than real history questions rated as good and mediocre, and real history questions rated as good have higher priority than real history questions rated as mediocre. When the ratings of classic excellence are equal, they are compared based on the number of historical uses, and the historical real questions with the smallest number of historical uses are selected; Establish multiple modification and adjustment models, and call one or more modification and adjustment models to modify and adjust historical questions to generate new questions according to needs. The modification and adjustment models include: selection order replacement model, conditional answer replacement model, numerical reset model and multi-question reorganization model; The selection order replacement model is used to replace the order of options in multiple-choice questions and generate new questions and new answers after the selection order is replaced; The conditional answer exchange model is used to exchange the given condition value and the answer value in the question. The answer value is given in the question, and the original answer value is used as the new condition value. The original condition value is used as the new answer value, and the original condition value is solved to vertically generate a new question and a new answer according to the new condition. The numerical reset model is used to obtain the calculation formula in the original historical real questions, randomly reset the original conditional values ​​when the requirements of the questions are met, input the reset conditional values ​​into the calculation formula, obtain the new answer results, and generate new questions and new answers based on the new conditional values; The multi-question reorganization model is used to combine multiple historical questions in the same knowledge test point. First, the calculation formula in the original historical question is obtained to obtain multiple historical questions that call the same calculation formula. The causal relationship between the condition value and the answer value in each historical question and the condition value and the answer value in other historical questions is determined. The condition values ​​and answer values ​​with the same meaning in multiple historical questions are modified and unified, and then the modified condition values ​​and answer values ​​are input into the calculation formula for calculation verification, and new questions and new answers are generated according to the verified condition values ​​and answer values. The test paper generation module is used to fill the selected historical real questions and generated new questions into the test paper template according to the constraints to generate the test paper.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for selecting questions and assembling examination papers based on big data and big models as described in any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for selecting questions and composing examination papers based on big data and big models as described in any one of claims 1 to 3 are implemented.

Images