Intelligent test question generation method based on large language model

Abstract

The invention discloses an intelligent test question generation method based on a large language model, and the method comprises the steps: obtaining a training courseware, and configuring question types needed by an examination and the number of test questions corresponding to each question type; generating a corresponding test question prompt word project according to a question type required by an examination, extracting courseware text information according to the obtained training courseware, configuring the courseware text information and the number of test questions corresponding to the question type as parameters to the corresponding test question prompt word project, generating a test question prompt word project to be debugged, and debugging the test question prompt word project to be debugged. Debugging the test question cue word project to be debugged; inputting the debugged test question prompt word engineering into the large language model in sequence, and generating test questions, answers and question analysis of the number of test questions corresponding to the question type required by the test; and storing the test questions, answers and question analysis of the test question quantity corresponding to the question type required by the test. The corresponding test paper can be quickly generated according to user requirements, the professional requirement for teaching and research personnel is reduced, the workload is further reduced, and the test question generation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent test question generation method based on a large language model. Background Art

[0002] In training and learning, examination and acceptance are an indispensable part. Whether it is for the education of primary and secondary school students, workplace training or the issuance of various qualification certificates, most of them need to design a comprehensive and effective examination based on the knowledge points involved. However, traditional examinations require teaching and research personnel to manually design examination questions based on the content of the training, which requires high professionalism of teaching and research personnel, and has a large workload and low efficiency. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide an intelligent test question generation method based on a large language model to improve the efficiency of test question design.

[0004] To solve the above technical problems, the purpose of the present invention is achieved through the following technical solutions: providing an intelligent test question generation method based on a large language model, comprising the following steps: obtaining training courseware, configuring the question types required for the test and the number of test questions corresponding to each question type; generating corresponding test question prompt word projects according to the question types required for the test, extracting courseware text information according to the obtained training courseware, configuring the courseware text information and the number of test questions corresponding to the question types as parameters to the corresponding test question prompt word projects, generating test question prompt word projects to be debugged, and debugging the test question prompt word projects to be debugged; inputting the debugged test question prompt word projects into the large language model in sequence, generating test questions, answers and question analyses corresponding to the number of test questions of the question types required for the test; storing the generated test questions, answers and question analyses corresponding to the number of test questions of the question types required for the test.

[0005] Its further technical solution is: the corresponding test question prompt word project is generated according to the question types required for the examination, the courseware text information is extracted according to the obtained training courseware, and the courseware text information and the number of test questions corresponding to the question types are configured as parameters to the corresponding test question prompt word project. The steps of generating the test question prompt word project to be debugged include: determining the role of the prompt word; extracting the courseware text information according to the obtained training courseware, setting the target of the prompt word according to the question types required for the examination combined with the courseware text information and the number of test questions corresponding to the question types; setting the workflow of the prompt word; determining the format of the output result; and generating the test question prompt word project to be debugged.

[0006] A further technical solution is as follows: the step of debugging the test prompt word project to be debugged is specifically as follows: debugging the test prompt word project to be debugged by adjusting keywords or adding key instructions.

[0007] Its further technical solution is: before inputting the debugged test prompt word project into the large language model in sequence, it also includes: training the large language model through an API interface connected to the open source large model.

[0008] Its further technical solution is: the step of storing the generated test questions, answers and question analyses of the number of test questions corresponding to the test types required for the examination is specifically: converting the text analysis corresponding to the number of test questions, answers and question analyses of the generated test questions, answers and question analyses of the number of test questions corresponding to the test types required for the examination into structured data, and storing the structured test questions, answers and question analyses of the number of test questions corresponding to the test types required for the examination in a database.

[0009] Its further technical solution is: after the step of storing the generated test questions, answers and question analysis corresponding to the number of test questions required for the test, it also includes: obtaining the received list of test participants and issuing corresponding test questions based on the list.

[0010] Its further technical solution is: after the step of obtaining the received list of candidates participating in the examination and issuing the corresponding examination questions according to the list, it also includes: comparing the answers and the test results of the candidates, calculating the examination scores, and analyzing the examination results based on the examination scores.

[0011] The beneficial technical effect of the present invention is as follows: the intelligent test question generation method based on the large language model of the present invention configures the test types and the number of test questions corresponding to each test type by acquiring training courseware; generates a corresponding test question prompt word project according to the test types required, extracts courseware text information according to the acquired training courseware, configures the courseware text information and the number of test questions corresponding to the test type as parameters to the corresponding test question prompt word project, generates a test question prompt word project to be debugged, and debugs the test question prompt word project to be debugged; inputs the debugged test question prompt word project into the large language model in sequence, generates test questions, answers and question analyses corresponding to the number of test questions of the test types required for the test; stores the generated test questions, answers and question analyses corresponding to the number of test questions of the test types required for the test, so as to quickly generate corresponding test papers according to user needs, reduce the professional requirements for teaching and research personnel, and reduce the workload, so that the generated test questions can more comprehensively cover test points and knowledge points, and improve the efficiency of test question generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0013] Figure 1 A schematic diagram of the flow of an intelligent test question generation method based on a large language model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0014] 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 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.

[0015] See also Figure 1 As shown, Figure 1 A flow chart of an intelligent test question generation method based on a large language model provided in an embodiment of the present invention, wherein the intelligent test question generation method based on a large language model comprises the following steps:

[0016] Step S10, obtaining training courseware, configuring the question types required for the exam and the number of questions corresponding to each question type; wherein the question types required for the exam include at least one of single-choice questions, multiple-choice questions, true-or-false questions, fill-in-the-blank questions and essay questions.

[0017] Step S20: Generate a corresponding test prompt word project according to the question type required for the test, extract the courseware text information according to the obtained training courseware, configure the courseware text information and the number of test questions corresponding to the question type as parameters to the corresponding test prompt word project, generate a test prompt word project to be debugged, and debug the test prompt word project to be debugged; wherein the prompt word project is based on language and optimization input to the large language model to guide the model to generate more accurate and more expected output results. The prompt word project includes the design of instructions, the provision of context and input data to guide the large language model to generate the number of test questions, answers and question analysis corresponding to the question type required for the test.

[0018] Step S30, input the debugged test prompt word project into the large language model in sequence, and generate test questions, answers and analysis of the number of test questions corresponding to the test question types required for the test;

[0019] Step S40, storing the generated test questions, answers and analysis of the number of test questions corresponding to the test question types required for the test.

[0020] The intelligent test question generation method based on the large language model obtains training courseware, configures the test types required for the test and the number of test questions corresponding to each question type; generates a corresponding test question prompt word project according to the test type required, extracts courseware text information according to the obtained training courseware, configures the courseware text information and the number of test questions corresponding to the question type as parameters to the corresponding test question prompt word project, generates a test question prompt word project to be debugged, and debugs the test question prompt word project to be debugged; inputs the debugged test question prompt word project into the large language model in sequence, generates test questions, answers and question analyses corresponding to the number of test questions of the test type required for the test; stores the generated test questions, answers and question analyses corresponding to the number of test questions of the test type required for the test, so as to quickly generate corresponding test papers according to user needs, reduce the professional requirements for teaching and research personnel, and reduce workload, so that the generated test questions can cover test points and knowledge points more comprehensively, and improve the efficiency of test question generation.

[0021] Specifically, the process of generating corresponding test prompt words according to the test question types required in step S20 includes:

[0022] Determine the role of the prompt word; wherein the role of the prompt word refers to the role of the test question corresponding to the question type required for the test and used to generate the test question of the corresponding question type. For example, when the question type required for the test is a true or false question, the role of the prompt word can be determined as a true or false question intelligent question generator.

[0023] The courseware text information is extracted based on the obtained training courseware, and the target of the prompt words is set according to the question types required for the examination combined with the courseware text information and the number of test questions corresponding to the question types; the target of the prompt words is to include keywords related to the question types required for the examination and the courseware text information. For example, when the question type required for the examination is a judgment question, the courseware text information and the number of test questions corresponding to the judgment question are combined to generate keywords for the judgment question related to the courseware text information for the number of test questions corresponding to the judgment question, and each question is in the form of being used to judge whether it is correct or incorrect.

[0024] Set the workflow of the prompt word to guide the large language model to generate output results according to the corresponding steps of the workflow of the corresponding prompt word;

[0025] Determine the format of the output results, wherein the output results include the number of questions corresponding to the question types required for the exam, answers, and question analyses; by determining the format of the output results, the output results are output in the specified format, thereby improving the accuracy of the output results and improving generalization.

[0026] Generate the test prompt word project to be debugged.

[0027] Specifically, the debugging of the test prompt word project to be debugged in step S20 is as follows:

[0028] Debug the test prompt word project to be debugged by adjusting keywords or adding key instructions.

[0029] Specifically, before step S30, the method further includes:

[0030] The big language model is trained through an API interface connected to an open source big model. Among them, the open source big model includes the Tongyi Qianwen big model or the Zhipu Qingyan big model. The corresponding open source big model can be provided through the API interface connected to the open source big model to debug the temperature value or parameter threshold of the training big language model, so as to improve the effect of test question generation by training the big language model.

[0031] Preferably, the step S40 is specifically:

[0032] The generated text analysis corresponding to the number of test questions, answers and analysis of the test questions of the required test question types is converted into structured data, and the structured test questions, answers and analysis of the number of test questions, answers and analysis of the test questions of the required test question types are stored in a database.

[0033] Specifically, after step S40, the method further includes:

[0034] Get the list of candidates who have taken the exam and send the corresponding exam questions according to the list.

[0035] Specifically, after the step of obtaining the received list of candidates participating in the examination and issuing corresponding examination questions according to the list, the step further includes:

[0036] Compare the answers and the test results of the examinees to calculate the test scores, and analyze the test results based on the test scores. Among them, the test scores can be statistically analyzed based on the dimensions of question type, question knowledge points, ranking, ranking percentage and / or whether the question is easy to make mistakes, and generate personal test reports and overall test reports.

[0037] In summary, the intelligent test question generation method based on a large language model of the present invention obtains training courseware, configures the question types required for the test and the number of test questions corresponding to each question type; generates a corresponding test question prompt word project according to the question types required for the test, extracts courseware text information according to the obtained training courseware, configures the courseware text information and the number of test questions corresponding to the question types as parameters to the corresponding test question prompt word project, generates a test question prompt word project to be debugged, and debugs the test question prompt word project to be debugged; inputs the debugged test question prompt word project into the large language model in sequence, generates test questions, answers and question analyses corresponding to the number of test questions of the question types required for the test; stores the generated test questions, answers and question analyses corresponding to the number of test questions of the question types required for the test, so that the corresponding test papers can be quickly generated according to user needs, reducing the professional requirements for teaching and research personnel, and reducing the workload, so that the generated test questions can cover test points and knowledge points more comprehensively, and improve the efficiency of test question generation.

[0038] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for generating intelligent test questions based on a large language model, characterized in that: The following steps are involved: Obtain training courseware, configure the question types required for the exam and the number of questions corresponding to each question type; Generate a corresponding test question prompt word project according to the question type required for the test, extract the courseware text information according to the obtained training courseware, configure the courseware text information and the number of test questions corresponding to the question type as parameters to the corresponding test question prompt word project, generate a test question prompt word project to be debugged, and debug the test question prompt word project to be debugged; Input the debugged test prompt word project into the large language model in sequence to generate the test questions, answers and analysis corresponding to the number of test questions required by the test; Store the questions, answers and analysis of the questions corresponding to the number of questions of the required question types for the generated exam.

2. The intelligent test question generation method based on a large language model according to claim 1, characterized in that: The steps of generating a corresponding test prompt word project according to the test type required for the test, extracting courseware text information according to the obtained training courseware, and configuring the courseware text information and the number of test questions corresponding to the test type as parameters to the corresponding test prompt word project, and generating the test prompt word project to be debugged include: Determine the role of the prompt word; Extract the courseware text information based on the obtained training courseware, and set the target of prompt words according to the required test questions, the courseware text information and the number of test questions corresponding to the test questions; Set the workflow for prompt words; Determine the format of the output results; Generate the test prompt word project to be debugged.

3. The intelligent test question generation method based on a large language model according to claim 1 is characterized in that: The steps for debugging the test prompt word project to be debugged are specifically as follows: Debug the test prompt word project to be debugged by adjusting keywords or adding key instructions.

4. The intelligent test question generation method based on a large language model according to claim 1, characterized in that: Before inputting the debugged test prompt word engineering into the large language model in sequence, the method further includes: Train large language models through API interfaces connected to open source large models.

5. The intelligent test question generation method based on a large language model according to claim 1 is characterized in that: The steps of storing the generated test questions, answers and analysis of the number of test questions corresponding to the test questions required by the test are specifically as follows: The generated text analysis corresponding to the number of test questions, answers and analysis of the test questions of the required test question types is converted into structured data, and the structured test questions, answers and analysis of the number of test questions, answers and analysis of the test questions of the required test question types are stored in a database.

6. The intelligent test question generation method based on a large language model according to claim 1, characterized in that: After the step of storing the generated test questions, answers and analysis of the number of test questions of the required test question types, the following step is further included: Get the list of candidates who have taken the exam and send the corresponding exam questions according to the list.

7. The intelligent test question generation method based on a large language model according to claim 6, characterized in that: After the step of obtaining the received list of candidates participating in the examination and issuing corresponding examination questions according to the list, the method further includes: Compare the answers with the test takers' test results, calculate the test scores, and analyze the test results based on the test scores.

Images