Automatic quality improvement method for science big question bank and related device

Abstract

The invention provides an automatic quality improvement method for a subject big question bank and a related device, and the method comprises the steps: obtaining a unit knowledge point list and a question sample of a specified unit, and constructing a cue word according to the question sample; taking the unit knowledge point list of the specified unit and the question sample as the context of a preset basic model, repeatedly calling a single question generation instruction according to the cue word, and generating a first question bank containing at least two questions; calling a question duplicate removal instruction to perform question duplicate removal on all questions in the first question bank, if repeated questions exist, calling a rewriting instruction to rewrite the repeated questions, and generating a second question bank without the repeated questions; and finally, calling a question quality inspection instruction to inspect and inspect the knowledge points of all questions in the first question bank or the second question bank and the answer accuracy so as to improve the question bank quality. According to the technical scheme, full-chain intellectualization from question construction to quality inspection can be realized, the construction cost is effectively reduced, and the question bank quality is improved.

Description

Technical Field

[0001] The present application relates to the field of information automation technology, and in particular to a method and related device for improving the automated quality of a science question bank. Background Art

[0002] With the development of information automation technology, the question bank management system in the field of education is gradually transforming towards intelligence. Especially in the field of science, there are many technologies dedicated to improving the automation of question bank construction and management, but it still faces many challenges, such as high costs caused by manual editing and low efficiency of manual error checking.

[0003] In the related technology, in terms of automatic evaluation of questions, the quality assessment has been carried out by using the similarity of answer results, but this method does not necessarily mean that the high similarity is the correctness of the answer, and it fails to fully guarantee the accuracy of the assessment. At the same time, in order to reduce repetitive work, it can be achieved by automatically generating template questions, but the cost of designing high-quality templates is high. In addition, the level of automation of question generation can be improved by introducing a large language model, but it is not enough when dealing with science questions that require precise calculation steps. Summary of the invention

[0004] In order to overcome the problems existing in the relevant technologies, the present application provides a method and related devices for automatically improving the quality of science question banks. In view of the particularity of science questions, by integrating automated question generation and quality control technology, the entire chain from question construction to quality inspection is intelligentized, effectively reducing construction costs and improving the quality of the question bank.

[0005] The first aspect of the present application provides a method for improving the automatic quality of a science question bank, comprising:

[0006] Obtain a unit knowledge point list and a question sample of a specified unit, and construct prompt words according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field, and a knowledge point field;

[0007] Using the unit knowledge point list of the designated unit and the example questions as the context of the preset basic model, repeatedly calling the single question generation instruction according to the prompt word to generate a first question bank including at least two questions;

[0008] Calling the question deduplication instruction to remove duplicate questions from all questions in the first question bank, and if there are duplicate questions, calling the rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions;

[0009] The question quality inspection instruction is called to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

[0010] The second aspect of the present application provides a system for improving the quality of a large science question bank automatically, comprising:

[0011] A first processing unit is used to obtain a unit knowledge point list and a question sample of a specified unit, and construct a prompt word according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field, and a knowledge point field;

[0012] A second processing unit is used to use the unit knowledge point list of the designated unit and the question sample as the context of the preset basic model, and repeatedly call the single question generation instruction according to the prompt word to generate a first question bank containing at least two questions;

[0013] A third processing unit is used to call a question deduplication instruction to remove duplicate questions from all questions in the first question bank, and if there are duplicate questions, call a rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions;

[0014] The fourth processing unit is used to call the question quality inspection instruction to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

[0015] A third aspect of the present application provides an electronic device, including:

[0016] Processor; and

[0017] A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method for automatically improving the quality of science question banks as described above.

[0018] A fourth aspect of the present application provides a non-transitory machine-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of an electronic device, the processor executes the method for automatically improving the quality of a science question bank as described above.

[0019] The technical solution provided by this application may have the following beneficial effects:

[0020] The technical solution of the present application provides a method and related device for improving the automated quality of a question bank for science questions, by obtaining a unit knowledge point list and a question sample of a specified unit, and constructing a prompt word based on the question sample, wherein the question sample includes at least a question field, an answer field, a step field, and a knowledge point field; using the unit knowledge point list of the specified unit and the question sample as the context of a preset basic model, repeatedly calling a single question generation instruction based on the prompt word to generate a first question bank containing at least two questions; calling a question deduplication instruction to dedupe all questions in the first question bank, and if there are duplicate questions, calling a rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions; calling a question quality inspection instruction to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank. The technical solution of the present application realizes the intelligence of the entire chain from question construction to quality inspection by integrating automated question generation and quality control technology, effectively reducing construction costs and improving the quality of question banks.

[0021] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other objects, features and advantages of the present application will become more apparent through a more detailed description of exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0023] Figure 1 It is a flow chart of a method for automatically improving the quality of a science question bank shown in an embodiment of the present application;

[0024] Figure 2 It is the result of executing the question deduplication stage shown in the embodiment of the present application;

[0025] Figure 3 It is a structural diagram of a system for improving the quality of science question bank automation shown in an embodiment of the present application;

[0026] Figure 4 It is a schematic diagram of the structure of an electronic device shown in an embodiment of the present application. DETAILED DESCRIPTION

[0027] The preferred embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0028] During the research, the applicant found that, especially for the particularity of science questions, how to realize the intelligence of the entire chain from question construction to quality inspection, effectively reduce the construction cost and improve the quality of the question bank is an urgent problem to be solved by technical personnel in this field.

[0029] In response to the above problems, the embodiments of the present application provide a method and related devices for automatically improving the quality of a science question bank. By integrating automated question generation and quality control technology, the entire chain from question construction to quality inspection is intelligentized, effectively reducing construction costs and improving the quality of the question bank.

[0030] The technical solution of the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0031] Figure 1 : This is a flow chart of a method for improving the quality of science question banks automatically according to an embodiment of the present application. The present application embodiment can be applied to different types of science questions, whether they are basic concept questions, application questions or complex comprehensive questions, and customized generation can be achieved by adjusting the prompt words and input context. The method can be performed by a system for improving the quality of science question banks automatically according to an embodiment of the present application, and the system can be integrated into various electronic devices such as computers and mobile phones.

[0032] See also Figure 1 The present application embodiment provides a method for improving the quality of a large science question bank automatically, and the method specifically comprises the following steps:

[0033] S101: Obtain a unit knowledge point list and a question sample of a specified unit, and construct prompt words according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field, and a knowledge point field.

[0034] The main principle of this application is to use the knowledge point list of a certain unit as context and generate the most critical content based on the question sample. This invention is mainly aimed at science questions, and requires that the generated content must include fields such as "question", "answer", "steps", and "knowledge point". This module combines the knowledge points of a certain science unit with the question sample, and uses the AIGC capability of the large-scale language model to generate questions and fill in fields.

[0035] It should be noted that this application requires that unit knowledge points be stored in the form of knowledge graphs and stored in the form of triples. The main sources of the graph are two aspects. One is manual construction based on human cognition of the unit content, such as self-compilation by subject teachers; the other is to combine natural language processing (NLP) technologies such as knowledge mining and graph construction for construction, and then review its correctness by professionals. This application needs to emphasize that the source of unit knowledge points must involve professionals to ensure its accuracy.

[0036] Furthermore, this application requires that the sample questions be described in structured text, and the structured methods include but are not limited to: json files, xml files or comma-delimited files that meet the format requirements. It should be noted that this application recommends that the selected structured format should be one that some large language models are good at generating, such as json format.

[0037] In the specific embodiment of the present application, the generation of 100 questions related to the first unit of "Description of Motion" of physics for the second year of high school is taken as an example to illustrate the implementation effect of the present solution.

[0038] First, in the preparation stage, the unit knowledge point list is stored in the form of a knowledge point map. Here, the construction process of the knowledge point map needs to be explained as follows:

[0039] First, we take the first unit of high school physics, "Description of Motion", as an example. The knowledge points involved in this unit include displacement, velocity, acceleration, etc. We construct a knowledge point map, involving three types of nodes, "course name, unit name, knowledge point name", and two types of relationships, "inclusion, association". The triple form of "(node, relationship, node)" is used as shown in Table 1:

[0040]

[0041] Table 1: Sample text of the knowledge graph used in the embodiments of this application, denoted as KG_STR

[0042] What is more important to note is that in order to ensure that all core concepts and their interrelationships are covered, this map is entered into the system after being reviewed and confirmed by the physics teacher team.

[0043] It should be noted that the knowledge graph is a replacement solution when the context is expressed as text: it can be replaced with other formats, for example, "(description of the movement, including, moment and time interval)" can be replaced with "<description of the movement, including, moment and time interval>", "[description of the movement, moment and time interval, including]", etc., as long as it does not affect the preset basic model's understanding of the knowledge point.

[0044] In addition, as a sample of structured questions, the embodiment of the present application selects a typical question as an example, such as "A car starts from rest and accelerates uniformly in a straight line. After 4 seconds, the speed reaches 20m / s. Find the acceleration." This question is converted into structured data represented in JSON format, and the sample question is shown in Table 2:

[0045]

[0046] Table 2: Examples of structured text used in the examples of this application

[0047] S102: Using the unit knowledge point list of the designated unit and the question sample as the context of the preset basic model, repeatedly calling a single question generation instruction according to the prompt word to generate a first question bank including at least two questions.

[0048] In this application, prompt words can be determined based on the sample questions, such as mainly including: clarifying the scope of knowledge points involved in the question based on the knowledge retrieval results; clarifying the format of the output content; requiring mathematical formulas to be expressed in standardized formats such as LaTeX or MathML, and to be compatible with the content format specified by the output content; requiring field names to be specified one by one according to the sample questions; and indicating one by one the fields that do not need to be filled in.

[0049] It should be noted that this application will select an open source large language model (such as the Tongyi Qianwen series) with good performance in code generation, context understanding, math problem solving and instruction execution as the preset basic model, and represent the knowledge graph and question examples in text as the context of the preset basic model. Finally, use appropriate prompt words to call the preset basic model to complete the generation of key information for a single question.

[0050] It should be emphasized that non-critical information (such as hash values, question numbers, etc.) should be generated in a parameterized manner in combination with predefined templates. In practical applications, multiple questions about the specified unit knowledge points can be generated by changing the unit knowledge points and repeatedly calling the single question generation instruction, thereby generating the first question bank. Further explanation, in the embodiment of the present application, the first question bank may contain repeated question types, or may not contain repeated question types.

[0051] In the specific embodiment of the present application, the prompt word is constructed as follows: prepare a string variable GRADE_STR = "second year of high school"; FIELD_STR = "id, original_text, HashCode, equation, ans, exam_point, procedure"; TYPE_STR = "physics".

[0052] According to the embodiment of the present application, prompt words are constructed. The prompt words in this embodiment are shown in Table 3. It should be noted that in actual use, string variables should be replaced with variable content texts.

[0053]

[0054] Table 3: Examples of prompt word texts containing string variables used in the title generation of the embodiment of the present application

[0055] It should be noted in the embodiments of the present application that string variables can be replaced in the process of constructing the question bank. For example, GRADE_STR can be replaced with "first grade of high school", TYPE_STR can be replaced with "mathematics", and KG_STR can be replaced with the knowledge graph text of the corresponding subject and unit.

[0056] In the specific embodiment of the present application, the basic model call can select Tongyi Qianwen Qwen2-72B-Instruct as the preset basic model, input the prompt word, and obtain the generated question information. It should be said that Tongyi Qianwen meets the requirements of the present application for better performance in code generation, context understanding, mathematical problem solving and instruction execution.

[0057] Further, in the output processing, the data returned by the question generation is parsed and automatically filled into the predefined template to generate the final question entry. It should be noted that the id field and type field have nothing to do with the question content and are filled in the template; the HashCode field will be uniformly calculated in the question deduplication module and is not filled in here, which can reduce the complexity of the preset basic model for instruction execution.

[0058] It should be noted that the above steps of "prompt word" construction, "preset basic model" call, and "output processing" are repeated until a sufficient number (100) of questions are generated. It should be noted that in the implementation scheme of this application, every 5 questions are saved into a separate json file, which can also be appropriately adjusted according to the actual file size, and is not specifically limited here.

[0059] S103: Call the question deduplication instruction to deduplicate all questions in the first question bank. If there are duplicate questions, call the rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions.

[0060] In the present application, the above-mentioned calling of the question deduplication instruction performs question deduplication on all questions in the first question bank, including:

[0061] Reading a text file corresponding to each question in the first question bank, and obtaining a question field, an answer field, and a step field corresponding to each question in the first question bank;

[0062] Concatenate the question field, answer field, and step field corresponding to each question to obtain a long string corresponding to each question;

[0063] Calculate the hash value of the long string corresponding to each question;

[0064] Traverse the hash values ​​of all long strings corresponding to the questions, check whether there are duplicate hash values, and record the duplicate questions.

[0065] Furthermore, in the present application, when there are repeated questions, a rewrite instruction is called to rewrite the repeated questions to generate a second question bank without repeated questions, including:

[0066] Reading questions with repeated hash values, taking any of the questions as a reference question, calling the rewriting instruction to construct a new prompt word, wherein the new prompt word includes the reference question and a rewriting instruction for the repeated question;

[0067] Based on the new prompt word, calling the single question generation instruction to generate a new question;

[0068] Calculate the hash value of the long character string corresponding to the new question until the second question bank without repeated questions is generated.

[0069] In this application, considering that the subsequent process needs to include question quality inspection, the deduplication in this application mainly checks whether the generated questions contain duplicate content from the question text and answer aspects. If duplicate questions are found, the following specific steps are performed, including:

[0070] First, concatenate the "title", "answer", and "step" fields of each question into a long string and calculate the hash value;

[0071] Secondly, traverse all the questions and check whether the hash values ​​are repeated;

[0072] Then, for questions with repeated hash values, one of the questions is used as a reference question, a prompt word is constructed, and the single question generation module is re-called to generate a new question.

[0073] This application emphasizes that the prompt words used in the last step, in addition to including reference questions, should include instructions for rewriting the original repeated questions, and explicitly require that the deduplication process adjust the question context, numerical settings or problem-solving paths while keeping the knowledge points unchanged, thereby generating new, differentiated question content. Further explanation: In the embodiment of this application, the second question bank does not contain repeated question types.

[0074] It should be noted that this de-duplication process can be designed as a loop mechanism until the generated question hash value has no duplication in the question bank.

[0075] In the specific embodiment of the present application, for each question, the steps of removing duplicate questions are as follows:

[0076] (1) Read the JSON file corresponding to the question and obtain the original_text (question), ans (answer), and procedure (steps) fields;

[0077] (2) Concatenate the original_text, ans, and procedure fields of each question into a long string, and use the MD5 algorithm to calculate the hash value of the string;

[0078] (3) Traverse all questions and check whether the hash values ​​are repeated; record all repeated questions.

[0079] (4) After checking, it was found that two of the questions had duplicate hash values. The execution results are as follows: Figure 2 As shown, Figure 2 It is the execution result of the title deduplication stage of the embodiment of the present application.

[0080] It should be noted that the calculation of the hash value can also be replaced by other text encoding methods, as long as different text encodings are unique and have the same length.

[0081] Therefore, in the embodiment of the present application, one of the questions is used as a reference question, prompt words are constructed, and the single question generation module is re-called to generate a new question.

[0082] Define the string variable JSON_STR as a reference title in json format, and construct the prompt words as shown in Table 4. It should be noted that in actual use, the string variables should be replaced with the variable content text.

[0083]

[0084] Table 4: Examples of prompt word texts containing string variables used in the title deduplication process in the embodiments of the present application

[0085] Furthermore, after obtaining a new question, the hash value is recalculated to complete the modification of the question. After the modification is completed, all questions are checked to confirm again that there are no repeated questions.

[0086] S104: Calling a question quality inspection instruction to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

[0087] In the present application, the above-mentioned calling question quality inspection instruction examines the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, including:

[0088] For each question in the first question bank or the second question bank, extract the corresponding question content text, answer content text, step content text and knowledge point content text;

[0089] Using the question content text and the knowledge point content text as context, verifying the accuracy of the knowledge point of each question;

[0090] When the accuracy of the knowledge points of the question is verified, the accuracy of the answer to each question is verified.

[0091] Furthermore, the above-mentioned using the question content text and the knowledge point content text as context to verify the accuracy of the knowledge point of each question includes:

[0092] Taking the question content text and the knowledge point content text as context, calling the preset basic model to determine whether the question content text contains the formula content text, if not, determining that the knowledge point of the question is inaccurate, and marking the corresponding question;

[0093] If so, it is determined that the knowledge points of the question are accurate.

[0094] Furthermore, the accuracy of the answers to each question is verified, including:

[0095] Based on the question content text, calling the preset basic model to solve the question, and obtaining the solution steps and answers of the corresponding question;

[0096] Comparing whether the answer given by the preset basic model is consistent with the answer provided by the question;

[0097] If the answers are consistent, then the solution steps are correct;

[0098] If the answers are inconsistent, extract the problem-solving steps and answers of the question, and determine whether there is an identical answer among the given answers;

[0099] If there are identical answers, the solution to the corresponding question will be any solution step corresponding to the identical answer as the new solution step;

[0100] If there is no identical answer, it is determined that the steps for solving the question are inaccurate, and the corresponding question is marked.

[0101] It should be further explained that in this application, the quality inspection process adopts an automated quality inspection technology solution based on reflection consistency, which aims to review the accuracy and rationality of the generated questions and their answers, so as to achieve the purpose of improving the quality of the question bank. The specific contents are as follows:

[0102] (1) Question information extraction: Take a question from the question bank and extract the content in the question, knowledge points, steps, and answer fields;

[0103] (2) Verification of the accuracy of question knowledge points: Take the question text and knowledge point text as context, call the basic model, and determine whether the question content involves the knowledge point text. If:

[0104] a) If the question content involves specified knowledge points, proceed to step (3)

[0105] b) If the question content does not involve the specified knowledge point, mark it and check the next question.

[0106] (3) Verification of answer accuracy. The specific plan is as follows:

[0107] 1) Based on the content of the question, call the basic model to solve the problem and obtain the solution steps and answers;

[0108] 2) Compare the answers given by the basic model with the answers provided in the question to see if they are consistent.

[0109] a) If the answers are consistent, the solution steps are considered correct and no changes are required to the question;

[0110] b) If the answers are inconsistent, let the basic model provide a new solution and a new answer, and determine whether there is an identical answer among the three answers.

[0111] i) If there are identical answers, the solution to this question will be any solution step corresponding to the identical answer as the new solution step;

[0112] ii) If the three answers are different, mark the question.

[0113] In this application, for all questions in the question bank that need quality inspection, repeat steps (1)-(3) to complete the quality inspection and quality improvement of the entire question bank.

[0114] It should be emphasized that for the marked questions, this application can consider deleting them from the question bank, or handing them over to the manual review team for verification and correction, which is not a necessary step for this solution.

[0115] In the specific implementation of this application, for each question, the following operations will be performed:

[0116] (1) Read the JSON file corresponding to the question and obtain the original_text (question), ans (answer), procedure (steps), and exam_point fields;

[0117] (2) Verification of the accuracy of the knowledge points in the questions:

[0118] a) Use prompt word 1 shown in Table 5 to call the preset basic model and get the answer of "yes" or "no". It should be noted that in actual use, the string variables should be replaced with the variable content text.

[0119]

[0120] Table 5: Examples of prompt word texts containing string variables used in the title quality inspection process in the embodiments of the present application

[0121] b) For questions with a “no” answer, proceed to step (3); otherwise, mark the question and check the next one.

[0122] (3) Verification of answer accuracy:

[0123] In the specific embodiment of the present application, based on the content of the question, the preset basic model is called using the prompt word 2 shown in Table 5 to solve the question and obtain the problem-solving steps and answers;

[0124] i) If the answers are consistent, the solution steps are considered correct and no changes are required to the question;

[0125] ii) If the answers are inconsistent, the preset basic model is asked to provide a new solution and a new answer through prompt word 2 shown in Table 5. At this time, the temperature parameter of the basic model should be set to ≥1 to ensure the diversity of generated answers.

[0126] For the 100 questions in the specific embodiment of the present application, the quality of the entire question bank can be improved by repeating steps (1)-(3).

[0127] It should be noted that the prompt words given in Tables 1 to 5 can be replaced with other expressions with the same semantics, and are not specifically limited here.

[0128] The embodiment of the present application provides a method for improving the automated quality of a question bank for science questions, by obtaining a unit knowledge point list and a question sample of a specified unit, and constructing a prompt word based on the question sample, wherein the question sample includes at least a question field, an answer field, a step field, and a knowledge point field; using the unit knowledge point list of the specified unit and the question sample as the context of a preset basic model, repeatedly calling a single question generation instruction based on the prompt word to generate a first question bank containing at least two questions; calling a question deduplication instruction to dedupe all questions in the first question bank, and if there are duplicate questions, calling a rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions; calling a question quality inspection instruction to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank. The technical solution of the embodiment of the present application integrates automated question generation and quality control technology to realize the full-chain intelligence from question construction to quality inspection, effectively reducing construction costs and improving the quality of the question bank.

[0129] Corresponding to the aforementioned application function realization method embodiment, the present application also provides a system for automated quality improvement of science question banks, an electronic device and corresponding embodiments.

[0130] Figure 3 Schematic diagram of the structure of the automatic quality improvement system for science question bank shown in the embodiment of the present application. Figure 3 The embodiment of the present application provides a system for improving the quality of a large science question bank automatically, and the system specifically includes:

[0131] The first processing unit 301 is used to obtain a unit knowledge point list and a question sample of a specified unit, and construct a prompt word according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field and a knowledge point field.

[0132] The second processing unit 302 is used to use the unit knowledge point list of the designated unit and the question sample as the context of the preset basic model, repeatedly call the single question generation instruction according to the prompt word, and generate a first question bank containing at least two questions.

[0133] The third processing unit 303 is used to call the question deduplication instruction to deduplicate all the questions in the first question bank, and if there are duplicate questions, call the rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions.

[0134] The fourth processing unit 304 is used to call the question quality inspection instruction to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

[0135] Furthermore, the third processing unit 303 is specifically configured to:

[0136] Reading a text file corresponding to each question in the first question bank, and obtaining a question field, an answer field, and a step field corresponding to each question in the first question bank;

[0137] Concatenate the question field, answer field, and step field corresponding to each question to obtain a long string corresponding to each question;

[0138] Calculate the hash value of the long string corresponding to each question;

[0139] Traverse the hash values ​​of all long strings corresponding to the questions, check whether there are duplicate hash values, and record the duplicate questions.

[0140] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0141] Figure 4 It is a schematic diagram of the structure of an electronic device shown in an embodiment of the present application.

[0142] See also Figure 4 , the electronic device 400 includes a memory 410 and a processor 420.

[0143] The processor 420 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0144] The memory 410 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. Among them, ROM can store static data or instructions required by the processor 420 or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory 410 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 410 may include a readable and / or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0145] The memory 410 stores executable codes, and when the executable codes are processed by the processor 420 , the processor 420 can execute part or all of the methods described above.

[0146] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0147] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program for initializing the following method steps:

[0148] Obtain a unit knowledge point list and a question sample of a specified unit, and construct prompt words according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field, and a knowledge point field;

[0149] Using the unit knowledge point list of the designated unit and the example questions as the context of the preset basic model, repeatedly calling the single question generation instruction according to the prompt word to generate a first question bank including at least two questions;

[0150] Calling the question deduplication instruction to remove duplicate questions from all questions in the first question bank, and if there are duplicate questions, calling the rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions;

[0151] The question quality inspection instruction is called to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

[0152] Furthermore, the calling of the question deduplication instruction to dedupe all questions in the first question bank includes:

[0153] Reading a text file corresponding to each question in the first question bank, and obtaining a question field, an answer field, and a step field corresponding to each question in the first question bank;

[0154] Concatenate the question field, answer field, and step field corresponding to each question to obtain a long string corresponding to each question;

[0155] Calculate the hash value of the long string corresponding to each question;

[0156] Traverse the hash values ​​of all long strings corresponding to the questions, check whether there are duplicate hash values, and record the duplicate questions.

[0157] Furthermore, when there are repeated questions, the rewriting instruction is called to rewrite the repeated questions to generate a second question bank without repeated questions, including:

[0158] Reading questions with repeated hash values, taking any of the questions as a reference question, calling the rewriting instruction to construct a new prompt word, wherein the new prompt word includes the reference question and a rewriting instruction for the repeated question;

[0159] Based on the new prompt word, calling the single question generation instruction to generate a new question;

[0160] Calculate the hash value of the long character string corresponding to the new question until the second question bank without repeated questions is generated.

[0161] Furthermore, the calling of the question quality inspection instruction to examine the accuracy of the knowledge points and answers of all the questions in the first question bank or the second question bank includes:

[0162] For each question in the first question bank or the second question bank, extract the corresponding question content text, answer content text, step content text and knowledge point content text;

[0163] Using the question content text and the knowledge point content text as context, verifying the accuracy of the knowledge point of each question;

[0164] When the accuracy of the knowledge points of the question is verified, the accuracy of the answer to each question is verified.

[0165] Furthermore, the question content text and the knowledge point content text are used as contexts to verify the accuracy of the knowledge points of each question, including:

[0166] Taking the question content text and the knowledge point content text as context, calling the preset basic model to determine whether the question content text contains the formula content text, if not, determining that the knowledge point of the question is inaccurate, and marking the corresponding question;

[0167] If so, it is determined that the knowledge points of the question are accurate.

[0168] Furthermore, the verification of the accuracy of the answer to each question includes:

[0169] Based on the question content text, calling the preset basic model to solve the question, and obtaining the solution steps and answers of the corresponding question;

[0170] Comparing whether the answer given by the preset basic model is consistent with the answer provided by the question;

[0171] If the answers are consistent, then the solution steps are correct;

[0172] If the answers are inconsistent, extract the problem-solving steps and answers of the question, and determine whether there is an identical answer among the given answers;

[0173] If there are identical answers, the solution to the corresponding question will be any solution step corresponding to the identical answer as the new solution step;

[0174] If there is no identical answer, it is determined that the steps for solving the question are inaccurate, and the corresponding question is marked.

[0175] The scheme of the present application has been described in detail above with reference to the accompanying drawings. In the above embodiments, the description of each embodiment has its own emphasis. For the part that is not described in detail in a certain embodiment, refer to the relevant description of other embodiments. Those skilled in the art should also know that the actions and modules involved in the description are not necessarily required for the present application. In addition, it can be understood that the steps in the method of the embodiment of the present application can be adjusted in order, merged and deleted according to actual needs, and the modules in the device of the embodiment of the present application can be merged, divided and deleted according to actual needs.

[0176] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0177] Alternatively, the present application can also be implemented as a non-temporary machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) on which executable code (or computer program, or computer instruction code) is stored. When the executable code (or computer program, or computer instruction code) is executed by a processor of an electronic device (or electronic device, server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0178] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the application herein may be implemented as electronic hardware, computer software, or combinations of both.

[0179] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system and method according to multiple embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a part of a module, a program segment or a code, and the part of the module, the program segment or the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or the flow chart, and the combination of the square boxes in the block diagram and / or the flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0180] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0181] It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0182] The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for improving the automated quality of a science question bank, characterized in that: include: Obtain a unit knowledge point list and a question sample of a specified unit, and construct prompt words according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field, and a knowledge point field; Using the unit knowledge point list of the designated unit and the example questions as the context of the preset basic model, repeatedly calling the single question generation instruction according to the prompt word to generate a first question bank including at least two questions; Calling the question deduplication instruction to remove duplicate questions from all questions in the first question bank, and if there are duplicate questions, calling the rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions; The question quality inspection instruction is called to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

2. The method according to claim 1, characterized in that The calling of the question deduplication instruction to dedupe all questions in the first question bank includes: Reading a text file corresponding to each question in the first question bank, and obtaining a question field, an answer field, and a step field corresponding to each question in the first question bank; Concatenate the question field, answer field, and step field corresponding to each question to obtain a long string corresponding to each question; Calculate the hash value of the long string corresponding to each question; Traverse the hash values ​​of all long strings corresponding to the questions, check whether there are duplicate hash values, and record the duplicate questions.

3. The method according to claim 2, characterized in that When there are repeated questions, the rewrite instruction is called to rewrite the repeated questions to generate a second question bank without repeated questions, including: Reading questions with repeated hash values, taking any of the questions as a reference question, calling the rewriting instruction to construct a new prompt word, wherein the new prompt word includes the reference question and a rewriting instruction for the repeated question; Based on the new prompt word, calling the single question generation instruction to generate a new question; Calculate the hash value of the long character string corresponding to the new question until the second question bank without repeated questions is generated.

4. The method according to claim 1, characterized in that: The calling of the question quality inspection instruction examines the accuracy of the knowledge points and answers of all the questions in the first question bank or the second question bank, including: For each question in the first question bank or the second question bank, extract the corresponding question content text, answer content text, step content text and knowledge point content text; Using the question content text and the knowledge point content text as context, verifying the accuracy of the knowledge point of each question; When the accuracy of the knowledge points of the question is verified, the accuracy of the answer to each question is verified.

5. The method according to claim 4, characterized in that The method of using the question content text and the knowledge point content text as context to verify the accuracy of the knowledge point of each question includes: Taking the question content text and the knowledge point content text as context, calling the preset basic model to determine whether the question content text contains the formula content text, if not, determining that the knowledge point of the question is inaccurate, and marking the corresponding question; If so, it is determined that the knowledge points of the question are accurate.

6. The method according to claim 4, characterized in that Verifying the accuracy of the answer to each question includes: Based on the question content text, calling the preset basic model to solve the question, and obtaining the solution steps and answers of the corresponding question; Comparing whether the answer given by the preset basic model is consistent with the answer provided by the question; If the answers are consistent, then the solution steps are correct; If the answers are inconsistent, extract the problem-solving steps and answers of the question, and determine whether there is an identical answer among the given answers; If there are identical answers, the solution to the corresponding question will be any solution step corresponding to the identical answer as the new solution step; If there is no identical answer, it is determined that the steps for solving the question are inaccurate, and the corresponding question is marked.

7. A system for improving the quality of science question banks, characterized in that: include: A first processing unit is used to obtain a unit knowledge point list and a question sample of a specified unit, and construct a prompt word according to the question sample, wherein the question sample at least includes a question field, an answer field, a step field, and a knowledge point field; A second processing unit is used to use the unit knowledge point list of the designated unit and the question sample as the context of the preset basic model, and repeatedly call the single question generation instruction according to the prompt word to generate a first question bank including at least two questions; A third processing unit is used to call a question deduplication instruction to remove duplicate questions from all questions in the first question bank, and if there are duplicate questions, call a rewrite instruction to rewrite the duplicate questions to generate a second question bank without duplicate questions; The fourth processing unit is used to call the question quality inspection instruction to review and inspect the accuracy of the knowledge points and answers of all questions in the first question bank or the second question bank, so as to improve the quality of the question bank.

8. The method according to claim 1, characterized in that The third processing unit is specifically used for: Reading a text file corresponding to each question in the first question bank, and obtaining a question field, an answer field, and a step field corresponding to each question in the first question bank; Concatenate the question field, answer field, and step field corresponding to each question to obtain a long string corresponding to each question; Calculate the hash value of the long string corresponding to each question; Traverse the hash values ​​of all long strings corresponding to the questions, check whether there are duplicate hash values, and record the duplicate questions.

9. An electronic device, characterized in that: include: processor; and a memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method for automatically improving the quality of science question banks as described in any one of claims 1-6.

10. A non-transitory machine-readable storage medium, characterized in that: An executable code is stored thereon, and when the executable code is executed by a processor of an electronic device, the processor is caused to execute the method for automatically improving the quality of a science question bank as described in any one of claims 1-6.

Images