Question and answer generation method and device based on knowledge graph, and automatic examination question generation system

Abstract

The application discloses a knowledge graph-based question and answer generation method, comprising the following steps: receiving a group question requirement of adding a question; using a question setting model to select corresponding knowledge samples from the current knowledge graph according to the knowledge points required by the group question requirement; and performing question type processing and organization on the knowledge samples to obtain a preliminary selected question. The application uses a question setting model to select corresponding knowledge samples from the current knowledge graph according to the knowledge points required by the group question requirement, and performs question type processing and organization on the knowledge samples to obtain a preliminary selected question. Since the examination questions are automatically generated by using the question setting model, the efficiency of manual question setting is obviously improved, and the accuracy of the generated questions is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent examination technology, specifically to a question-and-answer generation method based on knowledge graphs, a question-and-answer generation device based on knowledge graphs, an automatic question-and-answer generation system, an automatic exam question generation method, an electronic device, and a computer storage medium. Background Technology

[0002] In academic life, exams have become a necessary way for people to assess their professional skills. Exams for different subjects all require the basic steps of setting questions, compiling papers, and grading.

[0003] Currently, the process of creating and compiling exam questions is mostly done manually. However, considering the diversity of exam subjects and questions, the efficiency of manual question creation and compilation is relatively low. In addition, since both question creation and compilation are done manually, errors are inevitable in the generated exam questions, thus reducing the accuracy of the generated exam questions.

[0004] Furthermore, in many situations, a question-and-answer format for knowledge verification is necessary to enhance the engagement and effectiveness of knowledge transfer. Essentially, this is a form of examination, or rather, an examination is a specific question-and-answer format. The question is the one posing the questions, and the answer is the one being asked. When providing question-and-answer questions, it is essential to organize the questions according to the relevant knowledge points and prepare answers for verification.

[0005] Therefore, how to improve the efficiency of traditional manual question generation and test paper compilation, as well as the accuracy of generated questions and answers, has become an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0006] This application provides a knowledge graph-based question-and-answer generation method to address the existing problems of improving the efficiency and accuracy of traditional manual question generation and test paper compilation. This application also provides a knowledge graph-based question bank generation device, an automatic question-and-answer generation system, an automatic test question generation method, an electronic device, and a computer storage medium.

[0007] This application provides a question-answering generation method based on knowledge graphs, including:

[0008] Receive a request to add questions, which includes a question set requirement, and the question set requirement includes at least the knowledge points of the questions to be added;

[0009] Using the question-generating model, select the corresponding knowledge samples from the current knowledge graph according to the knowledge points required by the question-setting requirements;

[0010] The knowledge samples are processed and organized according to the question type to obtain the primary questions.

[0011] Optionally, the method comprises the following steps.

[0012] Receiving the authentication result of the primary questions;

[0013] According to the authentication result of the primary questions, the question setting model is trained and adjusted; and according to the artificial authentication result, the primary questions that pass the authentication are provided to the question bank.

[0014] Optionally, the knowledge graph is constructed by the following steps, comprising:

[0015] An initial knowledge graph and knowledge samples are obtained; the initial knowledge graph comprises knowledge labels and knowledge entities; the knowledge entities are directly or indirectly mounted under the knowledge labels representing specific knowledge fields according to the association between the information contained in the knowledge entities and the knowledge labels in a tree structure;

[0016] According to the information contained in each knowledge sample, the knowledge sample is mounted under the corresponding knowledge entity of the initial knowledge graph to form the current knowledge graph.

[0017] Optionally, the step of mounting the knowledge sample under the corresponding knowledge entity of the initial knowledge graph according to the information contained in the knowledge sample is realized by using a deterministic finite automaton algorithm.

[0018] Optionally, the tree structure is realized in the form of an inverted index.

[0019] Optionally, the method further comprises the following steps.

[0020] A request message for creating a special examination is obtained; the special examination corresponds to a specified knowledge field;

[0021] According to the request message, attribute information corresponding to the special examination is generated; the attribute information at least comprises an ID value of the examination and a key;

[0022] According to the requirements of the special examination in the request information, questions associated with the special examination are extracted from the question bank;

[0023] It is judged whether the extracted questions meet the requirements;

[0024] If yes, the attribute information of the special examination is added to the extracted questions, and the questions are stored in the examination question storage area;

[0025] If not, a request for adding questions to the question bank is proposed; the request for adding questions comprises the question setting requirements extracted according to the request message for creating a special examination.

[0026] Optionally, if a specific test paper needs to be created, the following steps are included:

[0027] receiving a test paper generation task, the test paper generation task including at least a question type requirement, a knowledge point requirement, and a number of questions requirement;

[0028] using the ID value of the specific test and the key, opening the question storage area, and reading the stored questions;

[0029] generating a corresponding test paper according to the question type requirement, the knowledge point requirement, and the number of questions requirement of the test paper generation task.

[0030] Optionally, the following steps are included:

[0031] adding examinee information to the test paper;

[0032] issuing the test paper to the examinee.

[0033] Optionally, the group question requirement includes one or more of the following requirements: a question type requirement, a number of questions requirement.

[0034] Optionally, the step of organizing the selected knowledge samples to obtain preliminary questions includes:

[0035] processing the knowledge samples that meet the conditions according to at least one of the following requirements: a question type requirement, a difficulty requirement, and a discrimination requirement, to obtain preliminary questions that meet the requirements.

[0036] The application also provides a knowledge graph-based question and answer generation device, which includes:

[0037] a question adding unit configured to receive a question adding requirement, the question adding requirement including a group question requirement, the group question requirement including at least a knowledge point of a question to be added;

[0038] a knowledge sample selection unit configured to use a question setting model to select corresponding knowledge samples from a current knowledge graph according to the knowledge points required by the group question requirement;

[0039] a preliminary question obtaining unit configured to process and organize the knowledge samples according to a question type to obtain preliminary questions.

[0040] The application also provides an automatic question and answer generation system, which includes:

[0041] a knowledge graph generation module configured to obtain an initial knowledge graph and knowledge samples, and to mount each knowledge sample to a corresponding knowledge entity of the initial knowledge graph according to information contained in the knowledge sample, so as to update the knowledge graph;

[0042] A question setting model is configured to select corresponding knowledge samples from the current knowledge graph according to group question requirements;

[0043] A question management module is configured to receive requests for adding, deleting and modifying questions, and to perform corresponding processing on the questions in the question bank according to the requests;

[0044] A knowledge point management module is configured to receive requests for adding, deleting and modifying knowledge points in the knowledge graph, and to perform corresponding processing on the knowledge labels in the knowledge graph according to the requests;

[0045] A group question management module is configured to receive group question requirements and to start a group question process according to the requirements.

[0046] Optionally, the application further comprises:

[0047] An examination question generation module is configured to receive request information for creating an examination, to generate attribute information corresponding to the examination according to the request information, and to include the attribute information of the special examination in the extracted questions and store the questions in the examination question storage area.

[0048] Optionally, the application further comprises:

[0049] A test paper generation module is configured to receive a test paper generation task, the test paper generation task including at least question type requirements, knowledge point requirements and examination question quantity requirements, to use the ID value and the key of the special examination to open the examination question storage area and read the stored questions, and to generate a corresponding test paper according to the question type requirements, knowledge point requirements and examination question quantity requirements in the test paper generation task.

[0050] The application further provides an automatic examination question generation method, comprising:

[0051] Obtaining request information for creating an examination task, the request information including at least group question requirements and special examination requirements;

[0052] Selecting corresponding knowledge samples from the current knowledge graph according to the group question requirements to generate preliminary selected questions;

[0053] Extracting questions associated with the special examination from the preliminary selected questions according to the special examination requirements to generate a test paper, the special examination requirements including at least question type requirements, question quantity requirements and corresponding score requirements;

[0054] Distributing the test paper.

[0055] The application further provides an electronic device, comprising a processor, and a memory for storing a computer program, which is run by the processor to execute the method of any one of the above.

[0056] The application further provides a computer storage medium, which stores a computer program, which is run by a processor to execute the method of any one of the above.

[0057] Compared with the prior art, the application has the following advantages:

[0058] The application provides a question and answer generation method based on a knowledge graph, comprising: receiving a requirement of adding examination questions, wherein the requirement of adding examination questions comprises a group question requirement, and the group question requirement comprises at least a knowledge point of an examination question to be added; using a question setting model to select a corresponding knowledge sample from a current knowledge graph according to the knowledge point required by the group question requirement; and performing type processing and organization on the knowledge sample to obtain a preliminary selected examination question. In the application, the question setting model is used to select a corresponding knowledge sample from a current knowledge graph according to the knowledge point required by the group question requirement, and the knowledge sample is subjected to type processing and organization to obtain a preliminary selected examination question. Since the examination question is automatically generated by using the question setting model, the efficiency of manual question setting is obviously improved. Moreover, the corresponding examination question is automatically generated by selecting a corresponding knowledge sample according to the knowledge point required by the group question requirement through the knowledge graph, so that the accuracy of the generated examination question is improved. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 FIG. 1 is a schematic diagram of an application scenario provided by the first embodiment of the application.

[0060] Figure 2 FIG. 2 is a flowchart of the question and answer generation method based on the knowledge graph provided by the first embodiment of the application.

[0061] Figure 3 FIG. 3 is a schematic diagram of the knowledge graph provided by the first embodiment of the application.

[0062] Figure 4 FIG. 4 is a schematic diagram of the automatic examination question generation system provided by the second embodiment of the application.

[0063] Figure 5 FIG. 5 is a schematic diagram of the question bank generation device based on the knowledge graph provided by the third embodiment of the application.

[0064] Figure 6 FIG. 6 is a flowchart of the automatic examination question generation method provided by the fourth embodiment of the application.

[0065] Figure 7A schematic diagram of an automatic examination question generation device provided by a fifth embodiment of the present application.

[0066] Figure 8 A schematic diagram of an electronic device provided by a fourth embodiment of the present application. DETAILED DESCRIPTION

[0067] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of embodiments of the present application. However, embodiments of the present application can be practiced in other ways not described in detail herein, and those skilled in the art can make similar extensions without departing from the connotations of embodiments of the present application, and therefore embodiments of the present application are not limited to the specific implementation disclosed below.

[0068] In order to enable those skilled in the art to better understand the scheme of the present application, the specific application scenarios of the embodiments of the customer image processing method provided by the present application are described in detail below, such as Figure 1 As shown in the figure, it is a schematic diagram of the application scenario corresponding to the knowledge graph-based question and answer generation method provided by the first embodiment of the present application.

[0069] The scene explains the knowledge graph-based question and answer generation method provided by the first embodiment of the present application with the scene of test takers taking exams, staff setting questions, and making test papers. In this scene, the generated questions can be called examination questions; in order to adapt to other non-examination scenarios, the examination questions can also be directly called questions. Among them, the test takers take exams using terminal devices, and the staff make test papers and set questions by submitting requests to the server side through the examination system client installed on the terminal device. The terminal device can be a desktop computer, a notebook computer, or a mobile phone, etc., and the desktop computer is preferred in this scene.

[0070] Specifically, for example, the subject that the test taker needs to take the exam is history, and for this exam subject, the staff setting questions and making test papers can determine the corresponding knowledge points, question type requirements, and the number of examination questions corresponding to each question type, wherein the question type requirements are, for example, single-choice question type, multiple-choice question type, judgment question type, and question and answer question type, which respectively correspond to 10 single-choice questions, 5 multiple-choice questions, 10 judgment questions, and 5 question and answer questions, etc. The staff inputs the test paper requirements, so that the (first) terminal device obtains the requirements for increasing examination questions and sends the requirements for increasing examination questions to the processor. The requirements for increasing examination questions include test paper requirements, and the test paper requirements at least include the knowledge points of the examination questions that need to be increased.

[0071] After the processor obtains the requirement of increasing the examination question, the question setting model is used to select corresponding knowledge samples from the current knowledge graph stored in the database of the server according to the knowledge points required by the question setting requirement, and the knowledge samples are processed and organized in the form of questions to obtain the preliminary examination question. The preliminary examination question can be directly presented on the display interface of the terminal device. At this time, the staff can authenticate the preliminary examination question to determine whether the preliminary examination question meets the examination requirements, for example, whether the knowledge points examined by the preliminary examination question are completely involved, whether the knowledge points exceed the scope, or whether the number of examination questions corresponding to each type of question of the preliminary examination question is correct. After the staff authentication is qualified, the server obtains the artificial authentication result for the preliminary examination question, which includes the qualified result and the unqualified result of the preliminary examination question. For the authentication result of the qualified preliminary examination question, the examination system provides the authenticated preliminary examination question to the question bank. At the same time, according to the authentication result of the qualified and unqualified preliminary examination question, the examination system trains and adjusts the question setting model on the server side to adjust the qualified question setting model, thereby serving the subsequent question setting.

[0072] In the above manner, the authenticated preliminary examination question is provided to the question bank, and after multiple rounds, a question bank with rich data and passed authentication is obtained, which can be used to automatically generate test papers.

[0073] The automatically generated test paper includes the following processes: the examination system obtains a request message for creating a special examination, and the request message provides at least a certain specified knowledge field corresponding to the special examination. Attribute information corresponding to the special examination is generated according to the request message; the attribute information at least includes the ID value of the examination and a key. According to the requirements of the special examination in the request message, the examination questions associated with the special examination are extracted from the question bank, and then it is determined whether the extracted examination questions meet the requirements. If so, the attribute information of the special examination is added to the extracted examination questions, and the examination questions are stored in the examination question storage area. After the corresponding examination questions are determined, the test paper for the examinee examination needs to be generated. Specifically, a test paper generation task is received, and the test paper generation task at least includes the question type requirement, the knowledge point requirement, and the examination question quantity requirement. The examination question storage area is opened using the ID value and the key of the special examination, and the stored examination questions are read. According to the question type requirement, the knowledge point requirement, and the examination question quantity requirement proposed by the test paper generation task, the corresponding test paper is generated. After the corresponding test paper is generated, the system obtains an examination request, adds examinee information to the test paper according to the examination request, and delivers the test paper to the terminal used by the examinee. The examination request is issued by the terminal device of the examinee, and when the specified time is reached, the examination system can deliver the test paper to the terminal device of the examinee according to the examination request, and the examinee can use the terminal device to take the examination.

[0074] It should be noted that in the present scenario, the examination system issues the test paper to the examinee, which can be based on the examinee directly sending an examination request to the server through the second terminal device to obtain the corresponding test paper, or indirectly sending an examination request to the first terminal device through the second terminal device, and then sending the corresponding test paper to the second terminal device used by the examinee after the first terminal device obtains the corresponding test paper from the processor.

[0075] The present scenario uses a question generation model to select corresponding knowledge samples from the current knowledge graph according to the required knowledge points of the question setting requirement, and processes and organizes the knowledge samples to obtain preliminary selected examination questions. Since the examination questions are automatically generated using the question generation model, the efficiency of manual question setting is significantly improved. In addition, the corresponding examination questions are automatically generated by selecting corresponding knowledge samples according to the required knowledge points of the question setting requirement through the knowledge graph, so that the accuracy of the generated examination questions is improved. In addition, the present scenario also manually authenticates the preliminary selected examination questions, further improving the accuracy of the automatically generated examination questions through the knowledge graph according to the question setting requirement.

[0076] Corresponding to the above scenario, the first embodiment of the present application provides a knowledge graph-based question and answer generation method, as shown in Figure 2 The flowchart of the knowledge graph-based question and answer generation method provided by the first embodiment of the present application is shown in Figure 2 Generally, Figure 2 The method is mainly executed on the server side with a database, and relevant requests are issued through a client that can access the server; of course, it is not excluded that an independent client is used for execution, and the client should have a large-capacity nonvolatile memory. The method includes the following steps:

[0077] Step S201, receiving a requirement for adding examination questions, wherein the requirement for adding examination questions includes a question setting requirement, and the question setting requirement at least includes a knowledge point of an examination question to be added.

[0078] In the first embodiment of the present application, when a test question required for an examination needs to be generated, a staff member or a learning system will determine the requirements of the examination question, and input the requirements of the examination question into the examination system, so that the examination system receives the requirement for adding examination questions, wherein the requirement for adding examination questions includes a question setting requirement, and the question setting requirement at least includes a knowledge point of an examination question to be added. The question setting requirement also includes one or more of the following requirements: a question type requirement, a number of examination questions requirement. Among them, the question type requirement is, for example, a single-choice question type, a multiple-choice question type, a judgment question type, and a question and answer question type, which respectively correspond to 10 single-choice questions, 5 multiple-choice questions, 10 judgment questions, and 5 question and answer questions, etc.

[0079] Step S202, using the question generation model, selecting corresponding knowledge samples from the current knowledge graph according to the required knowledge points of the group question requirements.

[0080] After receiving the requirement of increasing the test questions, using the question generation model, selecting corresponding knowledge samples from the current knowledge graph according to the required knowledge points of the group question requirements. In this step, the question generation model is generated by pre-training the knowledge sample of the knowledge graph, and the question generation model can be combined with the knowledge graph. The question generation model mainly includes the structure of the question type, such as single selection, multiple selection, judgment, and question and answer.

[0081] In this step, the knowledge graph is constructed by combining Figure 3 As shown in the figure, the knowledge graph is constructed by the following steps, including: first, obtaining an initial knowledge graph and a knowledge sample, wherein the initial knowledge graph contains knowledge labels and knowledge entities, and the knowledge entities are directly or indirectly mounted under the knowledge label representing a specific knowledge field in a tree structure according to the association relationship between the information contained by the knowledge entities and the knowledge labels. Among them, the tree structure is realized in the form of inverted index.

[0082] In the first embodiment of the present application, the association relationship between the information contained by the knowledge entities and the knowledge labels is determined by the following method: first, determine the association coefficient between the information contained by different knowledge entities and the knowledge labels, for example, the knowledge label is science, and the information contained by the knowledge entities includes food, computer research, XX (the name of a scientist), etc. Among them, the association coefficient between the food information and the knowledge label is 0.1, the association coefficient between the computer research information and the knowledge label is 0.6, and the association coefficient between the XX (the name of a scientist) information and the knowledge label is 0.5; then determine the association degree of the information of the computer research and the XX (the name of a scientist) knowledge entity. Then, the number of times of the different information appearing together within the predetermined range of the knowledge label is counted, and the association relationship between the information contained by the different knowledge entities and the knowledge labels is determined according to the number and the association degree.

[0083] In the first embodiment of the present application, the category of the knowledge label and the number of knowledge entities are both multiple, for example, one of the category names of the knowledge label is science, and among the numerous knowledge entities, there are XX character entities, space shuttle entities, wing entities, wheel group entities, etc. The knowledge entities associated with the knowledge label with the category name of science include space shuttle entities, wing entities, and wheel group entities, which are directly or indirectly mounted under the knowledge label representing a specific knowledge field in a tree structure. Of course, in the first embodiment of the present application, the category name of the knowledge label can also be other, for example, the category name of the knowledge label is history, the category name of the knowledge label is mathematics, and the category name of the knowledge label is literature, etc.

[0084] After obtaining the initial knowledge graph and the knowledge samples, according to the information contained in each knowledge sample, the knowledge sample is mounted under the corresponding knowledge entity of the initial knowledge graph to form the current knowledge graph. In the first embodiment of the present application, the step of mounting the knowledge sample under the corresponding knowledge entity of the initial knowledge graph according to the information contained in each knowledge sample is implemented by using a deterministic finite automaton algorithm. The deterministic finite automaton algorithm is a conventional algorithm in the art, and its specific content will not be explained here.

[0085] In step S203, the knowledge samples are processed and organized according to the question type to obtain the preliminary test questions.

[0086] After selecting the corresponding knowledge samples from the current knowledge graph according to the required knowledge points for the question setting, the knowledge samples are processed and organized according to the question type to obtain the preliminary test questions.

[0087] Specifically, the knowledge samples are processed and organized according to the question type in the question setting model to obtain the preliminary test questions. The question type structures in the question setting model include single-choice questions, multiple-choice questions, true or false questions, and essay questions. The knowledge samples are added to the corresponding question type according to the question type structure, and the knowledge samples that meet the requirements are processed according to at least one of the question requirements, difficulty requirements, and discrimination requirements to obtain the preliminary test questions that meet the requirements.

[0088] The first embodiment of the present application generates test questions based on the knowledge graph and the required knowledge points for question setting, which realizes the automatic generation of test questions. Compared with manual question setting and question setting, the efficiency is significantly improved. Moreover, since the test questions are automatically generated based on the knowledge graph and the required knowledge points for question setting, the accuracy of the generated test questions is improved. In addition, the knowledge graph covers a wide range of knowledge points, which increases the diversity of test questions and makes the distribution of knowledge points more balanced, avoiding the defects of biased questions and difficult questions.

[0089] After obtaining the preliminary test questions, in order to determine whether the preliminary test questions are qualified, i.e., whether the preliminary test questions meet the question setting requirements. Specifically, after obtaining the preliminary test questions, the preliminary test questions are presented on the display interface of the terminal device, and the staff will authenticate the preliminary test questions according to the set question setting requirements. The manual authentication result includes a qualified result and an unqualified result. The qualified result means that the preliminary test questions meet the question setting requirements, i.e., the preliminary test questions include the required knowledge points of the test questions; the unqualified result means that the preliminary test questions do not meet the question setting requirements, i.e., the preliminary test questions do not include the required knowledge points of the test questions. After the manual authentication is completed, the test system receives the manual authentication result of the preliminary test questions and executes the subsequent steps.

[0090] After receiving the artificial authentication result of the preliminary examination question, the artificial authentication result is used to train and adjust the question setting model according to the result of the artificial authentication result of the preliminary examination question, which is the result of passing or failing the preliminary examination question. That is, the artificial authentication result can be used as a sample for further training of the question setting model, wherein the authentication result is a positive sample if it is passed, and a negative sample if it is failed. For the result of passing the preliminary examination question, the preliminary examination question that passes the authentication is provided to the question bank.

[0091] Of course, in one way, the third-party authentication can also be used to determine whether the preliminary examination question is qualified. After the third-party authentication of the preliminary examination question, the corresponding authentication result is fed back to the examination system, and the examination system receives the third-party authentication result of the preliminary examination question and performs subsequent steps.

[0092] By continuously performing the above steps, the question bank continuously includes suitable questions. After the question bank has a sufficient number of questions, it can be selected according to the requirements when needed, provided to learners, especially provided with a test paper composed of questions.

[0093] The examination system will obtain a request message for creating a special examination, which comes from a certain learning system or a request from a certain client, and the special examination corresponds to a certain specified knowledge field. For example, the knowledge field is mathematics, or literature, or law. Then, attribute information corresponding to the special examination is generated according to the request message, wherein the attribute information at least includes the ID value of the examination and the key. And according to the requirements of the special examination in the request information, the examination questions associated with the special examination are extracted from the question bank. Then, it is determined whether the extracted examination questions meet the requirements in an artificial manner or in an automatic manner. If so, the attribute information of the special examination is added to the extracted examination questions, and the examination questions are stored in the examination question storage area.

[0094] After the corresponding examination question is determined, the test paper for the examinee examination needs to be generated. Specifically, a test paper generation task is received, which at least includes the type requirement, the knowledge point requirement and the number of examination questions requirement. The ID value and the key of the special examination are used to open the examination question storage area, read the stored examination questions, and generate the corresponding test paper according to the type requirement, the knowledge point requirement and the number of examination questions requirement of the test paper generation task. After the corresponding test paper is generated, the system will obtain an examination request, add examinee information to the test paper according to the examination request, and issue the test paper to the terminal corresponding to the examinee. The examination request is issued by the terminal device of the examinee. When the specified time is reached, the examination system can issue the test paper to the terminal device of the examinee according to the examination request, and the examinee can take the examination through the terminal device used by the examinee.

[0095] In addition, in the first embodiment of the present application, it is judged whether the extracted examination question meets the requirements, and if not, a request for adding an examination question is made to the question bank, and the request for adding an examination question includes the group question request extracted according to the request message for creating a special examination. According to the group question request, the specific step contents of steps S201-S203 are repeated.

[0096] In the first embodiment of the present application, in order to further improve the accuracy of question setting and group question setting in combination with the specific situation of the examinee examination, it further includes: first, information extraction is performed on the test paper to obtain examinee score information of each knowledge point in the test paper. Each knowledge point in a test paper is different, and the mastery degree of each examinee for each knowledge point is also different. Through analysis of the examinee score information of each knowledge point in the test paper, the mastery degree of the examinee for the knowledge point is determined. Specifically, each knowledge point in the test paper can be obtained, the number of examinees and the average score information of each examinee for each knowledge point are determined, and the average score information of each examinee for each knowledge point is used as the examinee score information of each knowledge point.

[0097] Then, the correlation between the knowledge points in the test paper and the obtained student score information is established, and the knowledge points required by the group question request are updated according to the established correlation, the question setting model is used to select the corresponding knowledge sample from the current knowledge graph according to the updated knowledge points required by the group question request, the knowledge sample is processed and organized in the form of a question, and the preliminary selected examination question is obtained. In this step, the correlation between the knowledge points in the test paper and the obtained student score information is established, including: taking the theme, key points, knowledge points contained in the test paper and the student score information as entities, and establishing the correlation between the entities according to the key points contained in the theme, the knowledge points contained in the key points and the student score information of each knowledge point.

[0098] The first embodiment of the present application reflects the mastery degree of the examinee for the corresponding knowledge point by combining the score of the examinee on the test question, evaluates the learning ability of the examinee for the knowledge point and the difficulty of the current test question, and performs group question setting in combination with the feedback information, thereby improving the efficiency and accuracy of question setting.

[0099] The first embodiment of the present application provides a question and answer generation method based on a knowledge graph, comprising: receiving a request to add a test question, wherein the request to add a test question includes a group question requirement, and the group question requirement at least includes a knowledge point of a test question that needs to be added; using a test question generation model to select a corresponding knowledge sample from a current knowledge graph according to the knowledge point required by the group question requirement; performing question type processing and organization on the knowledge sample to obtain a preliminary test question; receiving a manual authentication result of the preliminary test question; training and adjusting the test question generation model according to the manual authentication result of the preliminary test question; and providing the preliminary test question that passes the authentication to the question bank according to the manual authentication result. The first embodiment of the present application uses a test question generation model to select a corresponding knowledge sample from a current knowledge graph according to the knowledge point required by the group question requirement, and performs question type processing and organization on the knowledge sample to obtain a preliminary test question. The question type processing is to form a corresponding test question form according to the question type requirement, that is, to complete the conversion of the knowledge sample to the test question. The organization is to combine the test questions, for example, to form a group of ten multiple-choice questions for the examination of a certain knowledge point, or to evaluate the difficulty of a certain test question for the use in organizing a test paper. The test question generation model is implemented by using artificial intelligence such as machine learning. The model is trained with the samples provided by the manual review process, so that the test question generation model can use the continuously accumulated data to improve its intelligence, and the selection of the knowledge sample, the question type processing and the organization process of the test question generation model can be continuously optimized and improved. In addition, in possible implementation manners, the test question generation model can be responsible only for selecting a corresponding knowledge sample from a current knowledge graph according to the knowledge point required by the group question requirement, and the question type processing and the organization process can be performed by a special "rule-based group question" process.

[0100] Since the test questions are automatically generated by using the test question generation model, the efficiency is obviously improved compared with manual test question generation. In addition, the test questions are automatically generated by selecting a corresponding knowledge sample according to the knowledge point required by the group question requirement through the knowledge graph, so that the accuracy of the generated test questions is improved. In addition, the first embodiment of the present application also performs manual authentication on the preliminary test question, so as to further improve the accuracy of the test questions automatically generated according to the group question requirement through the knowledge graph.

[0101] The second embodiment of the present application provides an automatic test question generation system. The test question generation system can be part of a learning system, or the test question generation system is independent, and different learning systems (which can be represented as client learning software) call the test question generation system to measure the learning effect. As shown in FIG. 2, the test question generation system comprises a knowledge graph generation module, a test question generation model, a test question management module, a knowledge point management module and a group question management module. The specific functions of each module will be described below. Figure 4 ​

[0102] The knowledge graph generation module is configured to obtain an initial knowledge graph and knowledge samples, and mount each knowledge sample to a corresponding knowledge entity in the initial knowledge graph according to information contained in the knowledge sample, so as to update the knowledge graph.

[0103] Specifically, the initial knowledge graph contains knowledge labels and knowledge entities, and the knowledge entities are directly or indirectly mounted under the knowledge labels representing specific knowledge fields in a tree structure according to the association between the information contained in the knowledge entities and the knowledge labels. The tree structure is implemented in an inverted index manner.

[0104] In the second embodiment of the present application, the association between the information contained in the knowledge entities and the knowledge labels is determined by the following method. First, the association coefficients between the information contained in different knowledge entities and the knowledge labels are determined. For example, the knowledge label is science, and the information contained in the knowledge entities includes food, computer research, XX (the name of a scientist), and the like. The association coefficient between the food information and the knowledge label is 0.1, the association coefficient between the computer research information and the knowledge label is 0.6, and the association coefficient between the XX (the name of a scientist) information and the knowledge label is 0.5. Then, it is determined that the information of the knowledge entities of computer research and XX (the name of a scientist) has an association degree. Then, the number of times that different information appears together within a predetermined range of the knowledge label is counted, and the association between the information contained in different knowledge entities and the knowledge label is determined according to the number and the association degree.

[0105] In the second embodiment of the present application, the number of knowledge label categories and the number of knowledge entities are both multiple. For example, one of the knowledge label categories is science, and the XX person entity, the space shuttle entity, the wing entity, and the wheel group entity are included in the numerous knowledge entities. The knowledge entities that have an association with the knowledge label of the category name science include the space shuttle entity, the wing entity, and the wheel group entity, which are directly or indirectly mounted under the knowledge label representing the specific knowledge field in a tree structure. Of course, in the second embodiment of the present application, the category name of the knowledge label can also be other, such as the category name of the knowledge label being history, the category name of the knowledge label being mathematics, and the category name of the knowledge label being literature, and the like.

[0106] After obtaining the initial knowledge graph and the knowledge samples, each knowledge sample is mounted to a corresponding knowledge entity in the initial knowledge graph according to the information contained in the knowledge sample, to form the current knowledge graph. In the second embodiment of the present application, the step of mounting each knowledge sample to a corresponding knowledge entity in the initial knowledge graph according to the information contained in the knowledge sample is implemented by using a deterministic finite automaton algorithm. The deterministic finite automaton algorithm is a conventional algorithm in the field, and its specific content will not be explained here.

[0107] A question generation model is used to select corresponding knowledge samples from the current knowledge graph according to the group question requirement. Specifically, after obtaining the knowledge graph, the question generation model is used to select corresponding knowledge samples from the current knowledge graph according to the knowledge points required by the group question requirement. In this step, the question generation model is generated by pre-training of knowledge samples of the knowledge graph, and the question generation model can be combined with the knowledge graph. The question generation model mainly includes the following question types: single-choice, multiple-choice, true or false, and question and answer. The question generation model can be continuously trained by manual authentication or third-party authentication during use, that is, training while using, to gradually improve its intelligence. The specific model architecture can be implemented by various machine learning methods.

[0108] The group question requirement at least includes the knowledge points of the required added examination questions. The group question requirement further includes one or more of the following requirements: question type requirement, examination question quantity requirement. The question type requirement is, for example, single-choice, multiple-choice, true or false, and question and answer, which correspond to the following examination question quantities, respectively: 10 single-choice, 5 multiple-choice, 10 true or false, and 5 question and answer.

[0109] An examination question management module is used to receive requests for adding, deleting, and modifying examination questions, and accordingly process the examination questions in the question bank.

[0110] A knowledge point management module is used to receive requests for adding, deleting, and changing knowledge points in the knowledge graph, and accordingly process the knowledge labels in the knowledge graph.

[0111] A group question management module is used to receive group question requirements and accordingly start the group question process.

[0112] The automatic examination question generation system provided by the second embodiment of the present application further includes an examination question generation module configured to receive a request for creating an examination and generate attribute information of the examination according to the request information; the attribute information at least includes an ID value of the examination and a key; and according to the request for creating the examination, extract examination questions associated with the examination from the question bank, add the attribute information of the special examination to the extracted examination questions, and store the examination questions in an examination question storage area.

[0113] After the corresponding exam questions are determined, a test paper for the examinee to take the exam needs to be generated. The automatic exam question generation system further comprises a test paper generation module configured to receive a test paper generation task, wherein the test paper generation task at least comprises a question type requirement, a knowledge point requirement and a number of exam questions requirement; use the ID value and the key of the special exam to open the exam question storage area and read the stored exam questions; and generate a corresponding test paper according to the question type requirement, the knowledge point requirement and the number of exam questions requirement proposed by the test paper generation task.

[0114] After the corresponding test paper is generated, the system obtains an exam request, adds examinee information to the test paper according to the exam request, and distributes the test paper to a terminal corresponding to the examinee. The exam request is sent by the terminal of the examinee. When a specified time is reached, the exam system can distribute the test paper to the terminal of the examinee according to the exam request, and the examinee can take the exam by using the terminal.

[0115] The second embodiment of the present application provides an automatic exam question generation system, comprising: a knowledge graph generation module configured to obtain an initial knowledge graph and knowledge samples, and mount each knowledge sample to a corresponding knowledge entity of the initial knowledge graph according to information contained in the knowledge sample, so as to update the knowledge graph; a question generation model configured to select corresponding knowledge samples from the current knowledge graph according to a question setting requirement; an exam question management module configured to receive a request for adding, deleting or modifying an exam question, and perform corresponding processing on the exam question in the question bank according to the request; a knowledge point management module configured to receive a request for adding, deleting or changing a knowledge point in the knowledge graph, and perform corresponding processing on the knowledge label in the knowledge graph according to the request; and a question setting management module configured to receive a question setting requirement and start a question setting process according to the requirement. The automatic exam question generation system of the second embodiment of the present application selects corresponding knowledge samples from the knowledge graph according to the knowledge points required by the question setting requirement by using the question generation model, and processes and organizes the knowledge samples to obtain preliminary exam questions. Since the exam questions are automatically generated by using the question generation model, the efficiency of manual question setting is obviously improved. Moreover, the corresponding exam questions are automatically generated by selecting corresponding knowledge samples according to the knowledge points required by the question setting requirement through the knowledge graph, so that the accuracy of the generated exam questions is improved.

[0116] Corresponding to the knowledge graph-based question and answer generation method provided by the first embodiment of the present application, the third embodiment of the present application correspondingly provides a knowledge graph-based question and answer generation device. Since the device embodiment is basically similar to the first embodiment, the description is relatively simple, and the related parts are described in the part of the first embodiment. The device embodiment described below is only illustrative.

[0117] Please refer to Figure 5The third embodiment of the application provides a knowledge graph-based question bank generation device, comprising: an added examination question unit 501 configured to receive an added examination question requirement, wherein the added examination question requirement comprises a group question requirement, and the group question requirement comprises at least a knowledge point of an added examination question; a knowledge sample selection unit 502 configured to select a corresponding knowledge sample from a current knowledge graph according to the knowledge point required by the group question requirement by using a question setting model; a preliminary selected examination question obtaining unit 503 configured to obtain a preliminary selected examination question by subject type processing and organization of the knowledge sample; an artificial authentication result receiving unit 504 configured to receive an artificial authentication result of the preliminary selected examination question; a processing unit 505 configured to train and adjust the question setting model according to the artificial authentication result of the preliminary selected examination question; and provide the examination question bank with the authenticated preliminary selected examination question according to the artificial authentication result.

[0118] Optionally, the knowledge graph is constructed by the following steps: obtaining an initial knowledge graph and knowledge samples; the initial knowledge graph comprises knowledge labels and knowledge entities, and the knowledge entities are directly or indirectly mounted under the knowledge labels representing specific knowledge fields in a tree structure according to the association relationship between the information contained in the knowledge entities and the knowledge labels; and according to the information contained in each knowledge sample, the knowledge sample is mounted under the corresponding knowledge entity of the initial knowledge graph to form the current knowledge graph.

[0119] Optionally, the step of mounting each knowledge sample under the corresponding knowledge entity of the initial knowledge graph according to the information contained in the knowledge sample is implemented by using a deterministic finite automaton algorithm.

[0120] Optionally, the tree structure is implemented in an inverted index manner.

[0121] Optionally, the device further comprises an examination question generation unit configured to obtain a request message for creating a special examination, wherein the special examination corresponds to a specified knowledge field; generate attribute information corresponding to the special examination according to the request message; the attribute information at least comprises an ID value of the examination and a key; extract examination questions associated with the special examination from the question bank according to the requirements of the special examination in the request information; determine whether the extracted examination questions meet the requirements; if yes, add the attribute information of the special examination to the extracted examination questions, and store the examination questions in an examination question storage area; and if no, submit a requirement for adding examination questions to the question bank, wherein the requirement for adding examination questions comprises a group question requirement extracted according to the request message for creating a special examination.

[0122] Optionally, the test paper generation unit, used to create a test paper for a specific exam if necessary, includes: receiving a test paper generation task, the test paper generation task including at least question type requirements, knowledge point requirements, and number of test questions requirements; using the ID value and key of the specific exam, opening the test question storage area and reading the stored test questions; and generating the corresponding test paper according to the question type requirements, knowledge point requirements, and number of test questions requirements proposed by the test paper generation task.

[0123] Optionally, the test paper generation unit is also used to add candidate information to the test paper and distribute the test paper to the candidate.

[0124] Optionally, the requirements for setting up the questions may include one or more of the following: requirements for question type and requirements for the number of questions.

[0125] Optionally, the preliminary test question acquisition unit is specifically used to process knowledge samples that meet the requirements according to at least one of the requirements of question type, difficulty, and discrimination, so as to obtain preliminary test questions that meet the requirements.

[0126] The fourth embodiment of this application also provides a method for automatically generating exam questions, such as... Figure 6 As shown, Figure 6 This is a flowchart illustrating an automatic test question generation method provided in the fourth embodiment of this application. Generally, Figure 6 The method primarily executes on a server with a database, and sends relevant requests through a client that can access the server; however, it is also possible to use a separate client, which should have a large capacity of non-volatile memory. The method includes the following steps:

[0127] Step 601: Obtain request information for creating an exam task; the request information includes at least the requirements for setting up questions and the requirements for a specific exam.

[0128] In the fourth embodiment of the present application, when it is necessary to generate test questions required by an examination, the examination system obtains request information for creating an examination task. The request information can be sent to the examination system by a terminal device used by a test taker, or the request information can be input to the examination system by a staff member. The request information at least includes a question setting requirement and a requirement for a special examination. The question setting requirement at least includes a knowledge point that needs to be added. The question setting requirement also includes one or more of the following requirements: a question type requirement, a number of questions requirement. The question type requirement is, for example, a single-choice question type, a multiple-choice question type, a judgment question type, and a question and answer question type, which respectively correspond to a number of questions of 10 single-choice questions, 5 multiple-choice questions, 10 judgment questions, and 5 question and answer questions. The requirement for a special examination at least includes a question type requirement, a number of questions requirement, and a corresponding score requirement. Specifically, for example, 10 single-choice questions correspond to a set score of 20 points, 5 multiple-choice questions correspond to a set score of 15 points, 10 judgment questions correspond to a set score of 20 points, and 5 question and answer questions correspond to a set score of 45 points.

[0129] In step 602, according to the question setting requirement, a corresponding knowledge sample is selected from the current knowledge graph to generate a preliminary selected question.

[0130] After obtaining the request information for creating an examination task, according to the question setting requirement in the request message, a corresponding knowledge sample is selected from the current knowledge graph to generate a preliminary selected question. Specifically, using a question setting model, according to the knowledge point required by the question setting requirement, a corresponding knowledge sample is selected from the current knowledge graph. In this step, the question setting model is generated by pre-training the knowledge sample of the knowledge graph, and the question setting model can be combined with the knowledge graph. The question setting model mainly includes question type structures such as single-choice question type, multiple-choice question type, judgment question type, and question and answer question type. In this step, the construction process of the knowledge graph can be referred to the explanation and description of the first embodiment described above, which will not be repeated here.

[0131] After selecting the corresponding knowledge sample from the current knowledge graph according to the knowledge point required by the question setting requirement, the knowledge sample is processed and organized in a question type to obtain a preliminary selected question.

[0132] Specifically, using each question type structure in the question setting model, the knowledge sample is processed and organized in a question type to obtain a preliminary selected question. Each question type structure in the question setting model includes a single-choice question type, a multiple-choice question type, a judgment question type, and a question and answer question type. According to each question type structure, the knowledge sample is added to each question type. According to at least one of the question type requirement, the difficulty requirement, and the discrimination requirement, the knowledge sample that meets the requirements is processed to obtain a preliminary selected question that meets the requirements.

[0133] After obtaining the preliminary examination question, in order to determine whether the preliminary examination question is qualified, that is, whether the preliminary examination question meets the group question requirements. Specifically, after obtaining the preliminary examination question, the preliminary examination question is presented on the display interface of the terminal device, and the staff will authenticate the preliminary examination question according to the set group question requirements. The manual authentication result includes a qualified result and an unqualified result. The qualified result means that the preliminary examination question meets the group question requirements, that is, the preliminary examination question at least includes the knowledge points of the examination question that needs to be added. The unqualified result means that the preliminary examination question does not meet the group question requirements, that is, the preliminary examination question does not include the knowledge points of the examination question that needs to be added. After the manual authentication is completed, the examination system receives the manual authentication result of the preliminary examination question, and executes the subsequent steps.

[0134] After receiving the manual authentication result of the preliminary examination question, according to the manual authentication result of the preliminary examination question, the manual authentication result is an unqualified result of the preliminary examination question, and the out-of-question model is trained and adjusted according to the unqualified result. Or, according to the manual authentication result, the manual authentication result is a qualified result of the preliminary examination question, and the preliminary examination question that passes the authentication is provided to the question bank according to the qualified result.

[0135] Of course, in one way, the third-party authentication can also be used to determine whether the preliminary examination question is qualified. After the preliminary examination question is authenticated by the third party, the corresponding authentication result is fed back to the examination system, the examination system receives the third-party authentication result of the preliminary examination question, and executes the subsequent steps.

[0136] Through the continuous execution of the above steps, the question bank continuously includes suitable examination questions. After the question bank has a sufficient number of examination questions, the question bank can provide examination papers to learners when needed.

[0137] Step 603, extracting an examination question associated with the special examination from the preliminary examination question according to the requirements of the special examination to generate a test paper; the requirements of the special examination at least include a type requirement, a quantity requirement, and a corresponding score requirement.

[0138] After generating the preliminary examination question, an examination question associated with the special examination is extracted from the preliminary examination question according to the requirements of the special examination to generate a test paper. The special examination corresponds to a specified knowledge field. For example, the knowledge field is mathematics, literature, or law. Then, the attribute information corresponding to the special examination is determined, wherein the attribute information at least includes the ID value of the examination and the key. Then, according to the requirements of the special examination, an examination question associated with the special examination is extracted from the question bank storing the preliminary examination question. Then, it is judged whether the extracted examination question meets the requirements in a manual or automatic manner. If so, the attribute information of the special examination is added to the extracted examination question, and the examination question is stored in the examination question storage area.

[0139] After the corresponding exam questions are determined, a test paper for the examinee to take the exam is generated. Specifically, a test paper generation task is received, and the test paper generation task includes at least a question type requirement, a knowledge point requirement, and a number of exam questions requirement. Using the ID value of the special exam and the key, the exam question storage area is opened, the stored exam questions are read, and according to the question type requirement, the number of exam questions requirement, and the corresponding score requirement of the test paper generation task, a corresponding test paper is generated. For example, the generated test paper includes 10 single-choice questions, corresponding to a set score of 20 points, 5 multiple-choice questions, corresponding to a set score of 15 points, 10 judgment questions, corresponding to a set score of 20 points, and 5 essay questions, corresponding to a set score of 45 points.

[0140] In step 604, the test paper is issued.

[0141] After the corresponding test paper is generated, the system obtains an exam request, adds examinee information to the test paper according to the exam request, and issues the test paper to a terminal corresponding to the examinee. The exam request is issued by the terminal device of the examinee. When a specified time is reached, the exam system can issue the test paper to the terminal device of the examinee according to the exam request, and the examinee can take the exam by using the terminal device.

[0142] The fourth embodiment of the present application provides an automatic exam question generation method, including: obtaining request information of creating an exam task; the request information at least includes a group question requirement and a requirement of a special exam; generating preliminary selected exam questions from the current knowledge graph according to the group question requirement; extracting exam questions associated with the special exam from the preliminary selected exam questions to generate a test paper according to the requirement of the special exam; the requirement of the special exam at least includes a question type requirement, a number of exam questions requirement, and a corresponding score requirement; and issuing the test paper. The fourth embodiment of the present application generates preliminary selected exam questions from the current knowledge graph according to the knowledge points required by the group question requirement, and extracts exam questions associated with the special exam from the preliminary selected exam questions to generate a test paper according to the requirement of the special exam. Since the exam questions are automatically generated using the knowledge graph, the efficiency of manual question setting is obviously improved. Moreover, the corresponding exam questions are automatically generated by selecting the corresponding knowledge samples according to the knowledge points required by the group question requirement, so that the accuracy of the generated exam questions is improved.

[0143] Corresponding to the automatic exam question generation method provided by the fourth embodiment of the present application, the fifth embodiment of the present application correspondingly provides an automatic exam question generation device. Since the device embodiment is basically similar to the fourth embodiment, the description is relatively simple, and the related parts are described in the part of the fourth embodiment. The device embodiment described below is only illustrative.

[0144] Please refer toFigure 7 The fifth embodiment of this application provides an automatic exam question generation device, comprising: a request information obtaining unit 701, used to obtain request information for creating an exam task; the request information includes at least question grouping requirements and specific exam requirements; a preliminary question generation unit 702, used to select corresponding knowledge samples from the current knowledge graph according to the question grouping requirements to generate preliminary questions; an exam paper generation unit 703, used to extract questions related to the specific exam from the preliminary questions according to the specific exam requirements to generate an exam paper; the specific exam requirements include at least question type requirements, question quantity requirements, and corresponding score requirements; and an exam paper distribution unit 704, used to distribute the exam paper.

[0145] Corresponding to the knowledge graph-based question-answering generation method of the first embodiment of this application, the sixth embodiment of this application also provides an electronic device. For example... Figure 8 As shown, Figure 8 This is a schematic diagram of an electronic device provided in the sixth embodiment of this application. The electronic device includes: a processor 801; and a memory 802 for storing a computer program, which is executed by the processor to perform the knowledge graph-based question-and-answer generation method of the first embodiment and the automatic test question generation method provided in the fourth embodiment.

[0146] Corresponding to the knowledge graph-based question-answering generation method of the first embodiment and the automatic exam question generation method provided in the fourth embodiment of this application, the seventh embodiment of this application also provides a computer storage medium storing a computer program, which is executed by a processor to perform the knowledge graph-based question-answering generation method of the first embodiment and the automatic exam question generation method provided in the fourth embodiment.

[0147] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

[0148] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0149] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0150] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device. According to the definition herein, computer-readable media does not include transitory media, such as modulated data signals and carriers.

[0151] Those skilled in the art will appreciate that embodiments of the present application can be devised for a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, and the like) embodying computer-readable program code thereon for use by or in connection with an instruction execution system.

Claims

1. A question-answering generation method based on knowledge graphs, characterized in that, include: Receive a request to add questions, which includes a question set requirement, and the question set requirement includes at least the knowledge points of the questions to be added; Using a question-generating model, relevant knowledge samples are selected from the current knowledge graph according to the knowledge points required by the question-setting requirements. The knowledge graph is constructed through the following steps: obtaining an initial knowledge graph and knowledge samples; the initial knowledge graph contains knowledge tags and knowledge entities; the knowledge entities are directly or indirectly attached to the knowledge tags representing specific knowledge domains in a tree structure according to the association between their contained information and the knowledge tags; based on the information contained in each knowledge sample, it is attached to the corresponding knowledge entity in the initial knowledge graph to form the current knowledge graph. The knowledge samples are processed and organized into question types to obtain preliminary selection questions.

2. The question-answering generation method based on knowledge graphs according to claim 1, Other features include: Receive the authentication results for the preliminary selection of questions; Based on the authentication results of the preliminary selection questions, the question generation model is trained and adjusted; and based on the manual authentication results, the authenticated preliminary selection questions are provided to the question bank.

3. The question-answering generation method based on knowledge graphs according to claim 1, characterized in that, The step of attaching each knowledge sample to the corresponding knowledge entity in the initial knowledge graph based on the information contained in each knowledge sample is implemented using a deterministic finite automaton algorithm.

4. The question-answering generation method based on knowledge graphs according to claim 1, characterized in that, The tree structure is implemented using an inverted index.

5. The question-answering generation method based on knowledge graphs according to claim 1, characterized in that, Also includes: A request message is received to create a specialized exam; the specialized exam corresponds to a specific knowledge domain; Based on the request message, attribute information corresponding to the specific exam is generated; the attribute information includes at least the exam's ID value and a key; Based on the requirements for the specialized exam in the request information, questions related to the specialized exam are extracted from the question bank; Determine whether the selected questions meet the requirements; If so, the attribute information of the special examination will be added to the extracted questions, and the questions will be stored in the question storage area; If not, a request to add questions is made to the question bank; the request to add questions includes the question grouping requirements extracted based on the request message to create a special exam.

6. The question-answering generation method based on knowledge graphs according to claim 5, characterized in that, If you need to create a specific exam paper, then you will need to include: Receive a test paper generation task, which includes at least the requirements for question type, knowledge points, and number of questions; Using the ID value and key of the special exam, open the exam question storage area and read the stored questions; Based on the requirements for question types, knowledge points, and number of questions specified in the test paper generation task, generate the corresponding test paper.

7. The question-answering generation method based on knowledge graphs according to claim 6, characterized in that, include: Add candidate information to the exam paper; The test paper was distributed to the candidates.

8. The question-answering generation method based on knowledge graphs according to claim 1, characterized in that, The requirements for setting up questions include one or more of the following: requirements for question type and requirements for the number of questions.

9. The question-answering generation method based on knowledge graphs according to claim 1 or 7, characterized in that, The process of organizing the selected knowledge samples to obtain preliminary questions includes: Based on at least one of the requirements for question type, difficulty level, and discrimination, the knowledge samples that meet the conditions are processed to obtain preliminary selection questions that meet the requirements.

10. A question-answering generation device based on knowledge graphs, characterized in that, include: An additional question unit is provided for receiving requests to add questions. These requests include requirements for creating additional questions, which must include at least the knowledge points for the questions to be added. The knowledge sample selection unit is used to select corresponding knowledge samples from the current knowledge graph according to the knowledge points required by the question-setting requirements, using a question-generating model. The knowledge graph is constructed through the following steps: obtaining an initial knowledge graph and knowledge samples; the initial knowledge graph contains knowledge tags and knowledge entities; the knowledge entities are directly or indirectly attached to the knowledge tags representing specific knowledge domains in a tree structure according to the association between their contained information and the knowledge tags; and each knowledge sample is attached to the corresponding knowledge entity in the initial knowledge graph based on the information contained therein, thus forming the current knowledge graph. The preliminary question acquisition unit is used to process and organize the knowledge samples into question types to obtain preliminary questions.

11. An automatic question-and-answer generation system, characterized in that, include: The knowledge graph generation module is used to obtain an initial knowledge graph and knowledge samples, and to attach each knowledge sample to the corresponding knowledge entity of the initial knowledge graph according to the information contained in each knowledge sample, so as to update the knowledge graph. The question generation model is used to select appropriate knowledge samples from the knowledge graph according to the question requirements; The question management module is used to receive requests for adding, deleting, and modifying questions, and to process the questions in the question bank accordingly. The knowledge point management module is used to receive requests for adding, deleting, or modifying knowledge points in the knowledge graph, and to process the knowledge tags in the knowledge graph accordingly. The question-setting management module is used to receive question-setting requirements and initiate the question-setting process accordingly.

12. The automatic question-and-answer generation system according to claim 11, characterized in that, include: The exam question generation module is used to receive a request message to create an exam and generate attribute information corresponding to the exam based on the request message. The attribute information includes at least the ID value of the exam and a key; and, according to the request to create the exam, questions associated with the exam are extracted from the question bank, the attribute information of the exam is added to the extracted questions, and the questions are stored in the exam question storage area.

13. The automatic question-and-answer generation system according to claim 12, characterized in that, include: The test paper generation module is used to receive test paper generation tasks, which include at least the requirements for question types, knowledge points, and the number of test questions. Using the exam's ID value and key, open the exam question storage area and read the stored questions; generate the corresponding exam paper according to the question type requirements, knowledge point requirements, and number of questions required by the exam paper generation task.

14. An automatic test question generation method, characterized in that, include: Receive a request message to create an exam task; The request message shall include at least the requirements for setting up the questions and the requirements for the specific examination; According to the question-setting requirements, corresponding knowledge samples are selected from the current knowledge graph to generate preliminary questions; wherein, the knowledge graph is constructed through the following steps, including: obtaining an initial knowledge graph and knowledge samples; the initial knowledge graph contains knowledge tags and knowledge entities, and the knowledge entities are directly or indirectly attached to the knowledge tags representing specific knowledge domains in a tree structure according to the association between their contained information and the knowledge tags; according to the information contained in each knowledge sample, it is attached to the corresponding knowledge entity in the initial knowledge graph to form the current knowledge graph; According to the requirements of the specialized examination, questions related to the specialized examination are extracted from the preliminary selection of questions to generate an examination paper; the requirements of the specialized examination include at least the requirements for question type, the number of questions, and the corresponding score requirements; The test papers were distributed.

15. An electronic device, characterized in that, The electronic device includes: a processor; and a memory for storing a computer program, which is executed by the processor to perform the method according to any one of claims 1-9 and 14.

16. A computer storage medium, characterized in that, The computer storage medium stores a computer program that is executed by a processor to perform the method described in any one of claims 1-9 and 14.

Images