Medical theory examination intelligent test paper composition method based on artificial intelligence technology

Abstract

The invention discloses a medical theory examination intelligent test paper composition method based on an artificial intelligence technology, and the method comprises the following steps: constructing a medical question bank, determining an influence vector of each test question in the medical question bank, and constructing a plurality of sub-question banks based on the influence vectors; constructing a test paper structure model, optimizing the test paper structure model through multi-target constraint, extracting and caching test questions from the sub-question bank, and performing intelligent test paper composition on the cached test questions; checking and proofreading the test paper formed by the intelligent test paper composition, and performing secondary intelligent test paper composition on the cached test questions based on a proofreading result; through the intelligent test paper composition method, the influence of human factors on the test result is reduced, the fairness and authority of the test are ensured, and the system can carry out personalized adjustment according to the actual level and demand of the examinee, so that the test better meets the actual demand of the examinee.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent test paper composition, and in particular to an intelligent test paper composition method for medical theory examination based on artificial intelligence technology. Background Art

[0002] With the reform of information-based teaching and the development of artificial intelligence, online education and examination systems have become increasingly popular, and intelligent test paper generation technology has become one of the research hotspots in the field of teaching reform. However, the existing test paper generation methods have some obvious shortcomings: Limitations of manual test paper generation: The existing test paper generation method usually relies on teachers to manually generate the test paper, which is limited by the teachers' knowledge level and experience, and it is difficult to ensure the scientificity and fairness of the test paper. This method is not only time-consuming and labor-intensive, but also prone to duplication and imbalance of test paper content.

[0003] Irregular question bank management: Existing question bank management systems often lack effective classification and standardization, resulting in unscientific selection and extraction of test questions, making it difficult to meet the needs of different exams. The quality of test questions in the question bank varies, affecting the effectiveness of the exam.

[0004] Lack of personalization and adaptability: The existing test paper composition method cannot be personalized according to the test requirements, resulting in the test results not being able to truly reflect the candidates' knowledge mastery. Summary of the invention

[0005] The purpose of the present invention is to provide an intelligent test paper composition method for medical theory examinations based on artificial intelligence technology to solve the technical problems of the limitations of manual test paper composition, irregular question bank management, and lack of personalization and adaptability in the prior art.

[0006] In order to solve the above technical problems, the present invention specifically provides the following technical solutions: An intelligent test paper-forming method for medical theory examination based on artificial intelligence technology comprises the following steps: Constructing a medical question bank, determining an influence vector of each question in the medical question bank, and constructing a plurality of sub-question banks based on the influence vector; Constructing a test paper structure model and optimizing the test paper structure model through multi-objective constraints, extracting and caching test questions from the sub-question bank, and intelligently composing the cached test questions; The test papers generated by intelligent test paper compilation are checked and proofread, and based on the proofreading results, the cached test questions are intelligently compiled for the second time.

[0007] As a preferred solution of the present invention, test questions on medical theory examinations are collected and sorted; Classify and label the collected test questions, wherein the influence vector includes subject, knowledge point, difficulty level and test question number; A sub-question bank is constructed based on subject classification, and test questions of the same subject classification are stored in the same sub-question bank, and each test question in the sub-question bank is set with a unique matching corresponding test question number.

[0008] As a preferred embodiment of the present invention, Inputting examination requirements into the examination paper structure model, including the number of examination questions, the number of examination subjects, the knowledge point requirements, and the average level of examination difficulty; The test paper structure model extracts and caches test questions from the sub-question bank according to the input test requirements, and assigns corresponding scores to each test question based on the test question quantity requirements, test subject quantity requirements, knowledge point requirements, and test difficulty average level requirements.

[0009] As a preferred solution of the present invention, the test paper structure model is optimized by multi-objective constraints to select the sub-question bank, and cached test questions are extracted from the sub-question bank, wherein the specific implementation method of optimizing the test paper structure model by multi-objective constraints is: ; ; Where F(x) is the objective function vector; is the i-th objective function; x is a decision variable, including the question number of the question selected from each of the sub-question banks; the difficulty level of each question selected from each of the sub-question banks; the knowledge points covered by each selected question; the number of different difficulty levels corresponding to all selected questions; and the score corresponding to each selected question;

[0010] Ω is the examination requirement.

[0011] As a preferred solution of the present invention, the specific implementation method of optimizing the test paper structure model through multi-objective constraints is: Assign a weight to each objective function and synthesize multiple objective functions into a new objective function vector: ; in, is the weight of the i-th objective function, is the ith objective function.

[0012] As a preferred solution of the present invention, after the test paper structure model is used to intelligently organize the test questions selected for cache, the fitness of the test paper obtained by intelligently organizing the test questions each time is determined through fitness function analysis; The test paper composed of all test questions is regarded as an individual, each test question is mapped to a gene on the individual, the fitness value of each test paper individual is calculated using the fitness function, and the test paper individuals with high fitness values ​​are selected for reproduction to gradually optimize the quality of the test paper; The calculation formula of the fitness function is: ; in, The corresponding error is the test paper error caused by the i-th test paper composition factor on the intelligent test paper composition. is the weight coefficient; The test paper composition factors include the test paper knowledge point coverage, the test paper difficulty distribution coefficient, the number of test paper question types, the total score of each test paper and the test paper question repetition rate.

[0013] As a preferred embodiment of the present invention, test paper individuals with high fitness values ​​are selected for reproduction, and the specific reproduction method is: Determine the test questions contained in the test paper individuals with high fitness values; The test questions contained in the test paper individuals with high fitness values ​​are taken as the parent body, and the test questions corresponding to the parent body are selected again from the sub-question library through selection, crossover and mutation operations to generate new sub-test paper individuals.

[0014] As a preferred solution of the present invention, the method for checking and proofreading the test paper formed by intelligent test paper composition is as follows: Using fuzzy mathematics method, a fuzzy comprehensive evaluation model for test paper quality evaluation is established. The fuzzy comprehensive evaluation model is expressed as: ; Among them, B is the comprehensive evaluation result, W is the weight vector; R is the evaluation vector, and the evaluation vector is a multi-dimensional numerical vector, each dimension corresponds to a specific attribute or indicator of the test paper; Among them, the specific attributes corresponding to the multi-dimensional numerical vector include: The test paper’s knowledge point coverage, test paper difficulty distribution coefficient, number of test paper question types, total score of each test paper, and test paper question repetition rate.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention reduces the influence of human factors on the test results through an intelligent test paper composition method, ensures the fairness and authority of the test, and the system can be personalized according to the actual level and needs of the examinees, so that the test is more in line with the actual needs of the examinees. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0017] Figure 1 A schematic diagram of a flow chart of an intelligent test paper composition method according to an embodiment of the present invention; DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, the present invention provides an intelligent test paper composition method for medical theory examination based on artificial intelligence technology, comprising the following steps: Step 100: construct a medical question bank, determine the influence vector of each question in the medical question bank, and construct multiple sub-question banks based on the influence vector; In step 100, test questions about medical theory examination are collected and sorted; The collected test questions are classified and labeled, where the influence vectors include subject, knowledge point, difficulty level and test question number; A sub-question bank is constructed based on subject classification, and test questions of the same subject classification are stored in the same sub-question bank, and the test questions in each sub-question bank are set with unique matching corresponding test question numbers.

[0020] Step 200: construct a test paper structure model and optimize the test paper structure model through multi-objective constraints, extract and cache test questions from the sub-question bank, and intelligently organize the cached test questions; Input the examination requirements into the examination paper structure model, including the number of examination questions, the number of examination subjects, the knowledge point requirements and the average level of examination difficulty; The test paper structure model extracts and caches test questions from the sub-question bank according to the input test requirements, and assigns corresponding scores to each test question based on the test question quantity requirements, test subject quantity requirements, knowledge point requirements, and the average level of test difficulty.

[0021] The core function of the test paper structure model is to extract test questions that meet the requirements from the sub-question bank based on the input test requirements. To achieve this function, the model needs to receive the following input data: ① Required number of test questions: the total number of test questions that need to be included in the test paper.

[0022] ②Requirement on the number of examination subjects: the number of subjects that the examination paper needs to cover.

[0023] ③Knowledge point requirements: the range of knowledge points that the test paper needs to cover.

[0024] ④Average level of exam difficulty: the overall difficulty requirement of the test paper.

[0025] These input data constitute the constraints of the test paper structure model, and the model needs to extract test questions from the sub-question bank under these constraints.

[0026] When constructing a multi-objective constraint optimization model, the multi-objective constraint optimization algorithm is used to realize the extraction of test questions. Among them, the objective function vector F(x) of the test paper structure model contains multiple sub-objective functions, each sub-objective function corresponds to a test paper factor, and the objective function vector can be expressed as: ; in, represents the i-th objective function, including: knowledge point coverage (ensuring that the test paper covers all important knowledge points); difficulty distribution value (ensuring that the difficulty distribution of the test paper meets the examination requirements); question type distribution value (ensuring that the test paper contains a variety of question types); total score of the test paper (ensuring that the total score of the test paper meets the examination requirements); question repetition rate (ensuring that the questions in the test paper have a certain novelty and avoid repetition).

[0027] x is a decision variable, and x∈Ω, including the question number of the question selected from each sub-question bank; the difficulty level of each question selected from each sub-question bank; the knowledge points covered by each selected question; the number of different difficulty levels corresponding to all selected questions; and the score corresponding to each selected question.

[0028] Ω is a constraint condition, which includes the scope of the question bank (the questions in the test paper must come from the question bank); the test syllabus (the test paper must meet the requirements of the test syllabus); the number of questions (the number of questions in the test paper must be within a certain range); the difficulty of the questions (the difficulty distribution of the test paper must meet the requirements of the test); and the distribution of question types (the number of different question types in the test paper must meet the requirements of the test).

[0029] The specific implementation method of test question extraction based on the multi-objective constraint optimization model is: Step 1: Initialization: The model initializes the objective function and constraints based on the input exam requirements. For example, if the exam requires covering 5 knowledge points and the difficulty level is medium, the model will initialize the corresponding objective function and constraints.

[0030] Step 2: Question screening: The model selects questions that meet the initial constraints from the sub-question bank. For example, the model selects questions with a medium difficulty level and that cover the specified knowledge points.

[0031] Step 3: Multi-objective optimization: The model optimizes the selected test questions through multi-objective optimization algorithms (such as genetic algorithm, particle swarm optimization, etc.). The optimization goal is to minimize the objective function vector F(x), that is, to meet the requirements of all test paper composition factors as much as possible.

[0032] For example, the model will generate multiple test paper individuals (i.e., multiple test question combinations) through the selection, crossover, and mutation operations of the genetic algorithm, and calculate the fitness value of each test paper individual. The higher the fitness value, the more the test paper individual meets the test requirements.

[0033] Step 4: Question caching: The model caches the optimized test questions for subsequent intelligent test composition. The cached test questions include the test question number, difficulty level, knowledge point coverage, score and other information.

[0034] This implementation solves the problem of question duplication, ensures the uniqueness of each test paper, avoids the repetition of test questions, and increases the fairness and scientificity of the test paper. This is achieved through the diversity preservation mechanism in the intelligent test paper algorithm, ensuring the novelty and difference of each test question in the test paper. Furthermore, the specific implementation method of optimizing the test paper structure model through multi-objective constraints is: Assign a weight to each objective function and synthesize multiple objective functions into a new objective function vector: ; in, is the weight of the i-th objective function, is the ith objective function.

[0035] Step 300: Check and proofread the test paper formed by the intelligent test paper compilation, and conduct a second intelligent test paper compilation on the cached test questions based on the proofreading result.

[0036] After the test questions are extracted, the model will intelligently assemble the test questions based on the cached test questions. The specific steps are as follows: Step 1: Question distribution: The model assigns a corresponding score to each question based on the exam requirements. For example, if the total score required for the exam is 100 points, the model will reasonably assign a score to each question based on the difficulty of the question and the knowledge points covered.

[0037] Step 2: Test paper generation: The model arranges and combines the test questions with assigned scores according to certain rules (such as question type distribution, difficulty distribution, etc.) to generate the final test paper.

[0038] Step 3: Test paper inspection: The model verifies and proofreads the generated test paper through the fuzzy comprehensive evaluation model. The contents of the test include the coverage of test paper knowledge points, test paper difficulty distribution coefficient, number of test paper question types, total test paper score, and test paper question repetition rate. If the comprehensive evaluation result of the test paper is lower than the set threshold, the model will perform a second intelligent test paper until the generated test paper meets the requirements.

[0039] Assume the exam requirements are as follows: Number of questions: 50.

[0040] Examination subjects: Internal Medicine and Surgery.

[0041] Knowledge requirements: covering cardiovascular system, respiratory system, and digestive system.

[0042] Average level of exam difficulty: Medium.

[0043] The workflow of the test paper structure model is as follows: The model selects questions with a medium difficulty level from the sub-question banks of internal medicine and surgery, and these questions cover knowledge points of the cardiovascular system, respiratory system, and digestive system.

[0044] The model generates multiple test paper individuals through a multi-objective optimization algorithm and calculates the fitness value of each test paper individual.

[0045] The model selects the test paper individual with the highest fitness value and caches the test questions.

[0046] The model assigns scores to each question according to the exam requirements and generates the final test paper.

[0047] The model verifies and proofreads the generated test papers through a fuzzy comprehensive evaluation model to ensure that the test papers meet the examination requirements.

[0048] The implementation method of checking and proofreading the test paper formed by intelligent test paper composition is as follows: Using fuzzy mathematics method, a fuzzy comprehensive evaluation model for test paper quality evaluation is established. The fuzzy comprehensive evaluation model is expressed as: ; Among them, B is the comprehensive evaluation result, W is the weight vector; R is the evaluation vector, and the evaluation vector is a multi-dimensional numerical vector, each dimension corresponds to a specific attribute or indicator of the test paper; Among them, the specific attributes corresponding to the multi-dimensional numerical vector include: test paper knowledge point coverage, test paper difficulty distribution coefficient, the number of test paper question types, the total score of each test paper, and the test paper question repetition rate.

[0049] In the intelligent test-taking system, the evaluation vector is a set of key indicators used to measure the quality of the test paper. It reflects the performance of the test paper in multiple dimensions and is an important basis for optimizing the intelligent test-taking algorithm.

[0050] The evaluation vector is a multidimensional numerical vector, and each dimension corresponds to a specific attribute or indicator of the test paper: Specifically include: Knowledge point coverage: measure the degree of match between the knowledge points covered in the test paper and the knowledge points required by the test syllabus; Test paper difficulty distribution coefficient: evaluates whether the overall difficulty of the test paper meets expectations; Number of question types in the test paper: Ensure that the test paper contains a variety of question types to meet the examination requirements; Total score of the test paper: ensure that the total score of the test paper meets the examination requirements; Test question repetition rate: evaluate the novelty of questions in the test paper and avoid repetition; The application of evaluation vectors in the intelligent test paper system is mainly reflected in the following aspects: Optimize test paper quality: Through evaluation vectors, the system can comprehensively evaluate the quality of test papers and ensure that the test papers meet the examination requirements in multiple dimensions; Guiding the test paper generation process: The evaluation vector provides optimization direction for the intelligent test paper generation algorithm, helping the algorithm to select better test paper individuals; Feedback and Improvement: The results of the evaluation vector can provide feedback to the test paper designers, helping them understand the strengths and weaknesses of the test paper and make targeted improvements.

[0051] Based on the test paper structure model, the test paper formed by each intelligent assembly of the selected test questions is checked and proofread to obtain a comprehensive evaluation result B. According to the relationship between the comprehensive evaluation result B and the set threshold, the evaluation result corresponding to the test paper formed by the test paper structure model this time is determined. If the comprehensive evaluation result B is lower than the set threshold, a secondary intelligent assembly is required. When the comprehensive evaluation result B meets the set threshold, the fitness value corresponding to the test paper formed by each intelligent assembly of the selected test questions by the test paper structure model is determined, and test paper individuals with relatively large fitness values ​​are selected for reproduction. The adaptive genetic algorithm is used to simulate the biological evolution process, and the optimal solution is found through parallel search, adaptability and diversity maintenance, thereby improving the efficiency and quality of assembly.

[0052] After the test paper structure model is used to intelligently organize the test questions selected for cache, the fitness of the test paper obtained by intelligently organizing the test questions each time is determined through fitness function analysis; The test paper composed of all test questions is regarded as an individual, each test question is mapped to a gene on the individual, the fitness value of each test paper individual is calculated using the fitness function, and the test paper individuals with high fitness values ​​are selected for reproduction to gradually optimize the quality of the test paper; The calculation formula of the fitness function is: ; in, The corresponding error is the test paper error caused by the i-th test paper composition factor on the intelligent test paper composition. is the weight coefficient; The factors for test paper composition include the coverage of test paper knowledge points, the test paper difficulty distribution coefficient, the number of test paper question types, the total score of each test paper and the repetition rate of test paper questions.

[0053] When determining the fitness function for each test paper, the error value Indicates the deviation between the i-th test factor and the expected target. For each test factor, the error value The calculation method is as follows: The error of test paper knowledge point coverage on the test paper composed of intelligent test paper : Target value: coverage of knowledge points required by the examination syllabus.

[0054] Actual value: the proportion of knowledge points actually covered in the test paper.

[0055] Error value: =|Target knowledge point coverage - Actual knowledge point coverage| The error of the test paper difficulty distribution coefficient on the test paper composed of intelligent test paper : Target value: The difficulty distribution required for the exam (such as medium difficulty).

[0056] Actual value: actual difficulty distribution of the test paper.

[0057] Error value: =|Target Difficulty Distribution - Actual Difficulty Distribution| The error of the number of test questions on the test paper composed of intelligent test papers : Standard value: The number of question types required by the exam (such as multiple-choice questions, fill-in-the-blank questions, short-answer questions, etc.).

[0058] Actual value: the actual number of question types in the test paper.

[0059] Error value: =|Target number of questions - Actual number of questions| The error of the total score of the test paper to the test paper composed of intelligent test paper : Target value: the total score required for the exam (e.g. 100 points).

[0060] Actual value: the actual total score of the test paper.

[0061] Error value: =|Target total score - actual total score|.

[0062] The error of test paper composed of intelligent test paper due to the repetition rate of test questions : Target value: the test question repetition rate required by the exam (such as 0%).

[0063] Actual value: the actual repetition rate of test questions.

[0064] Error value: =|target repetition rate - actual repetition rate|.

[0065] Assume that the exam requirements are as follows: knowledge point coverage: 90%; difficulty distribution coefficient: medium; number of question types: 3; total score: 100 points; question repetition rate: 0%; The actual values ​​of the test paper generated at a certain time are: knowledge point coverage: 85%; difficulty distribution coefficient: medium to easy; number of question types: 2; total score: 95 points; question repetition rate: 5%; The error value is calculated as follows: =|90% - 85%| = 5% =|Medium - Medium Easy| = 0.2 (the difficulty distribution coefficient is a value between 0 and 1) =|3- 2| = 1 =|100 - 95| = ​​5 =|0% - 5%| = 5%.

[0066] Through the error values ​​corresponding to each of the above factors, the fitness function values ​​corresponding to the test papers generated by multiple intelligent test papers are calculated, and the test paper individuals with high fitness values ​​are selected for reproduction. The specific reproduction method is as follows: Determine the test questions contained in the test paper individuals with high fitness values; The test questions contained in the test paper individuals with high fitness values ​​are taken as the parent body, and the test questions corresponding to the parent body are selected again from the sub-question library through selection, crossover and mutation operations to generate new sub-test paper individuals.

[0067] The present invention reduces the impact of human factors on test results through an intelligent test paper composition method, ensures the fairness and authority of the test, and can make personalized adjustments based on the actual level and needs of the examinees, making the test more in line with the actual needs of the examinees. It also uses an adaptive genetic algorithm to simulate the biological evolution process, and finds the optimal solution through parallel search, adaptability and diversity maintenance, thereby improving the efficiency and quality of test paper composition.

[0068] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. An intelligent test paper-forming method for medical theory examination based on artificial intelligence technology, characterized in that: The following steps are involved: Constructing a medical question bank, determining an influence vector of each question in the medical question bank, and constructing a plurality of sub-question banks based on the influence vector; Constructing a test paper structure model and optimizing the test paper structure model through multi-objective constraints, extracting and caching test questions from the sub-question bank, and intelligently composing the cached test questions; The test papers generated by intelligent test paper compilation are checked and proofread, and based on the proofreading results, the cached test questions are intelligently compiled for the second time.

2. According to claim 1, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: Collect and organize questions about medical theory examinations; Classify and label the collected test questions, wherein the influence vector includes subject, knowledge point, difficulty level and test question number; A sub-question bank is constructed based on subject classification, and test questions of the same subject classification are stored in the same sub-question bank, and each test question in the sub-question bank is set with a unique matching corresponding test question number.

3. According to claim 1, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: Inputting examination requirements into the examination paper structure model, including the number of examination questions, the number of examination subjects, the knowledge point requirements, and the average level of examination difficulty; The test paper structure model extracts and caches test questions from the sub-question bank according to the input test requirements, and assigns corresponding scores to each test question based on the test question quantity requirements, test subject quantity requirements, knowledge point requirements, and test difficulty average level requirements.

4. According to claim 3, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: The test paper structure model is optimized by multi-objective constraints to select the sub-question bank and extract cached test questions from the sub-question bank, wherein the specific implementation method of optimizing the test paper structure model by multi-objective constraints is: ； ； Where F(x) is the objective function vector; is the i-th objective function; x is a decision variable, including the question number of the question selected from each of the sub-question banks; the difficulty level of each question selected from each of the sub-question banks; the knowledge points covered by each selected question; the number of different difficulty levels corresponding to all selected questions; and the score corresponding to each selected question; Ω is the examination requirement.

5. According to claim 4, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: The specific implementation method of optimizing the test paper structure model through multi-objective constraints is: Assign a weight to each objective function and synthesize multiple objective functions into a new objective function vector: ； in, is the weight of the i-th objective function, is the ith objective function.

6. According to claim 4, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: After the test paper structure model is used to intelligently organize the test questions selected for the cache, the fitness of the test paper obtained by intelligently organizing the test questions each time is determined through fitness function analysis; The test paper composed of all test questions is regarded as an individual, each test question is mapped to a gene on the individual, the fitness value of each test paper individual is calculated using the fitness function, and the test paper individuals with high fitness values ​​are selected for reproduction to gradually optimize the quality of the test paper; The calculation formula of the fitness function is: ； in, The corresponding error is the test paper error caused by the i-th test paper composition factor on the intelligent test paper composition. is the weight coefficient; The test paper composition factors include the test paper knowledge point coverage, the test paper difficulty distribution coefficient, the number of test paper question types, the total score of each test paper and the test paper question repetition rate.

7. According to claim 6, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: Select test paper individuals with high fitness values ​​for reproduction. The specific reproduction method is: Determine the test questions contained in the test paper individuals with high fitness values; The test questions contained in the test paper individuals with high fitness values ​​are taken as the parent body, and the test questions corresponding to the parent body are selected again from the sub-question library through selection, crossover and mutation operations to generate new sub-test paper individuals.

8. According to claim 1, a medical theory test intelligent test paper composition method based on artificial intelligence technology is characterized in that: The implementation method of checking and proofreading the test paper formed by intelligent test paper composition is as follows: Using fuzzy mathematics method, a fuzzy comprehensive evaluation model for test paper quality evaluation is established. The fuzzy comprehensive evaluation model is expressed as: ； Among them, B is the comprehensive evaluation result, W is the weight vector; R is the evaluation vector, and the evaluation vector is a multi-dimensional numerical vector, each dimension corresponds to a specific attribute or indicator of the test paper; Among them, the specific attributes corresponding to the multi-dimensional numerical vector include: The test paper’s knowledge point coverage, test paper difficulty distribution coefficient, number of test paper question types, total score of each test paper, and test paper question repetition rate.

Images