Test question answering method and device, electronic equipment and storage medium

Abstract

This invention provides a method, apparatus, electronic device, and storage medium for answering test questions. The method includes: acquiring test question data to be answered; determining prompts corresponding to the test question data based on the test question attributes and the user-oriented cognitive attributes of the test question data; and inputting the test question data and prompts into a large language model to obtain answer data for the test question data output by the large language model. The method, apparatus, electronic device, and storage medium provided by this invention enhance the relevance and adaptability of the prompts corresponding to the test question data in terms of both the test question data and the user's cognitive attributes, based on the test question attributes and the user-oriented cognitive attributes of the test question data. By inputting the resulting prompts into a large language model, more suitable test question attributes and user cognitive attributes can be obtained, thereby improving the quality of test answering, making the answer data easier for users to understand and accept, and enhancing the learning assistance effect.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a test question answering method, apparatus, electronic device, and storage medium. Background Technology

[0002] In homework and Q&A scenarios, intelligent learning devices can automatically explain the solution process and results of test questions through artificial intelligence technology, thereby tutoring students' learning.

[0003] However, current AI-based automated test-taking solutions often produce answer data that is difficult for students to understand or that does not match the test questions well, directly affecting the effectiveness of assisted learning. Summary of the Invention

[0004] This invention provides a test question answering method, apparatus, electronic device, and storage medium to address the shortcomings of automated test question answering outputs in related technologies that do not meet expectations.

[0005] This invention provides a method for answering test questions, including: Obtain the data of unanswered test questions; Based on the test question attributes of the test question data and the cognitive attributes of the test question data towards users, determine the prompts corresponding to the test question data; The test question data and the prompts are input into a large language model to obtain the answer data of the test question data output by the large language model.

[0006] According to a test question answering method provided by the present invention, determining the prompt corresponding to the test question data based on the test question attributes and the user-oriented cognitive attributes of the test question data includes: From the candidate prompt templates, select the prompt template that matches the question attributes of the question data; The cognitive attributes of the test questions are filled into the prompt template to obtain the prompt corresponding to the test questions.

[0007] According to a test question answering method provided by the present invention, the prompt template includes a description text of the answering steps for the test question data under the test question attribute.

[0008] According to a test question answering method provided by the present invention, the answering step description text includes a knowledge point extraction step description text, which describes the steps of extracting the knowledge points required to answer the test question data from knowledge points that conform to the cognitive attributes.

[0009] According to the test question answering method provided by the present invention, it further includes: Based on the user's registration information and / or the user's historical learning records, the cognitive attributes of the user's test data are determined.

[0010] According to a test question answering method provided by the present invention, the test question attributes include subject and / or question type.

[0011] The present invention also provides a test question answering device, comprising: The acquisition unit is used to acquire the test question data to be answered; The prompt construction unit is used to determine the prompt corresponding to the test question data based on the test question attributes and the user-oriented cognitive attributes of the test question data. The solution unit is used to input the test question data and the prompts into a large language model to obtain the solution data of the test question data output by the large language model.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement any of the above-described test question answering methods.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the test question answering method as described above.

[0014] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the test question answering method as described above.

[0015] The test-answering method, apparatus, electronic device, and storage medium provided by this invention enhance the relevance and adaptability of the prompts corresponding to the test data in terms of both the test data and the user's cognitive attributes, based on the test data's test-answering attributes. By inputting the resulting prompts into a large-scale language model, more suitable test-answering attributes and user cognitive attributes can be obtained, thereby improving the quality of test-answering, making the answer data easier for users to understand and accept, and enhancing the learning effect. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the test question answering method provided by the present invention.

[0018] Figure 2 This is a schematic diagram of the test question answering device provided by the present invention.

[0019] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] All actions involving the acquisition of signal information or data in this invention are carried out in compliance with the relevant data protection laws and policies of the country where the device is located, and with the authorization granted by the owner of the device.

[0022] With the development of artificial intelligence technology, large language models (LLMs) are being applied in educational settings.

[0023] Taking homework tutoring as an example, large-scale language models can be deployed on smart learning devices such as learning machines, dictionary pens, and scanning pens, or on the cloud accessible to these devices. Students can use these devices to photograph or scan test questions, and the large-scale language model will automatically generate the solution process and results. Thus, smart learning devices can use large-scale language models to automate the explanation of test solutions and results, thereby tutoring students.

[0024] However, in practical applications, automated test-taking solutions based on large-scale language models still have some problems.

[0025] Specifically, students of different ages have varying cognitive levels. However, current automated test-solving solutions do not consider these differences in cognitive level, often using the same set of prompts to generate both the solution process and the result. For example, when solving problems like the "chicken and rabbit in the same cage" problem using a large language model, the model might output a solution by setting up a system of equations. This approach represents advanced knowledge beyond the curriculum for elementary school students. Although the intelligent learning device provides explanations, students cannot truly grasp the solution logic from these explanations, directly impacting the effectiveness of automated test-solving.

[0026] Furthermore, in test-taking scenarios involving different question types, the same set of prompts is often used, which frequently results in insufficient adaptability to different question types. Insufficient adaptability of prompts directly affects the output quality of large-scale language models, potentially leading to incorrect results or outputting test-taking solutions that do not conform to the actual problem-solving approach.

[0027] For example, when answering cloze tests using generic prompts, the large language model's output might only omit the explanations for some sub-questions due to insufficient adaptability of the prompts to the specific cloze test. Specifically, when answering a cloze test with 15 blanks, the large language model's step-by-step solution might only include explanations for questions 2, 5, 7, and 14, omitting the explanations for the remaining 11 sub-questions.

[0028] To address the aforementioned problems, embodiments of the present invention provide a test question answering method. This method can be applied to intelligent learning devices, such as learning machines, dictionary pens, scanning pens, and character recognition pens, as well as other types of intelligent devices, such as smartphones and tablets. The embodiments of the present invention do not impose specific limitations on this method.

[0029] Figure 1 This is a flowchart illustrating the test question answering method provided by the present invention, as shown below. Figure 1 As shown, the method includes: Step 110: Obtain the test data to be answered.

[0030] Here, "question data" refers to the data related to the questions to be answered, specifically the question stem data, that is, the content of the questions to be answered. Question data can represent data used to directly display the specific content of the questions to the user, such as at least one of the following: the text description of the questions to be answered, formulas in the questions to be answered, and charts.

[0031] Specifically, the test data to be answered can be various types of test questions, such as math multiple-choice questions, math calculation questions, Chinese composition questions, Chinese reading comprehension questions, English cloze tests, English composition questions, etc. This embodiment of the invention does not make specific limitations on this.

[0032] The test data can be actively entered by the user. Specifically, the user can enter the test data by taking a picture, scanning, or importing a file. Alternatively, the user can enter the test data by swiping, clicking, or other touch operations, keyboard input operations, voice input operations, etc. This embodiment of the invention does not make specific limitations on this.

[0033] Step 120: Based on the question attributes of the question data and the user-oriented cognitive attributes of the question data, determine the prompts corresponding to the question data.

[0034] Specifically, for the acquired test question data to be answered, its test question attributes can be determined. Here, test question attributes refer to the attribute information reflected by the test question data, which may include the subject to which the test question data belongs, such as Chinese, mathematics, English, chemistry, physics, etc.; it may also include the question type to which the test question data belongs, such as fill-in-the-blank questions, multiple choice questions, true / false questions, application questions, proof questions, essay questions, etc.; it may also include the textbook version, chapter, and knowledge points tested by the test question data, etc., but this embodiment of the invention does not specifically limit these aspects.

[0035] Furthermore, for the acquired test question data to be answered, the cognitive attributes of the test question data towards the target user can be determined. Here, the target user of the test question data refers to the user who needs the explanation or answer of the test question data, that is, the user to whom the answer to the test question data is intended. In some embodiments, the target user of the test question data is the user who entered the test question data. The cognitive attributes of the test question data towards the target user are used to reflect the cognitive level or cognitive stage of the user to whom the test question data is intended. Cognitive attributes can usually be reflected by the user's academic year information, for example, the cognitive attribute can be 5th grade, 7th grade, or 10th grade, or preschool, etc.

[0036] It is understandable that the question attributes of test data can specifically reflect the characteristics of the test data itself. The cognitive attributes of the test data, which are geared towards the user, can specifically reflect the user's cognitive level. In this embodiment of the invention, the prompt required for answering the test question can be determined by combining the question attributes and the user-oriented cognitive attributes of the test data.

[0037] Here, the prompt is used during the input phase of a large language model to specify the requirement for answering test data; that is, the prompt instructs the large language model to generate answer data for the test data. Furthermore, in this embodiment of the invention, the prompt, determined based on the test data's test attributes and the user-oriented cognitive attributes of the test data, can concretely describe the adaptive answer requirements for test attributes and cognitive attributes when specifying the requirement for answering test data. Compared to traditional general prompts, the resulting prompts can more specifically describe, in natural language, the thought processes and rules to be followed when answering test data under a certain type of test attribute, and more specifically incorporate the user's cognitive attributes to be considered when answering test data, thereby enhancing the prompts' relevance and adaptability to both test data and user characteristics.

[0038] Step 130: Input the test question data and the prompts into a large language model to obtain the answer data of the test question data output by the large language model.

[0039] Specifically, after receiving the prompt, the prompt and the test data can be input into a large language model.

[0040] Large-scale language models here refer to Natural Language Processing (NLP) models with a large number of parameters, where the number of parameters and / or the complexity of the model structure exceed a preset threshold. These models process large-scale text data during training and possess the ability to understand and generate natural language. Examples of large-scale language models include the Spark Large Model.

[0041] Furthermore, the large language model here can be a general-domain large language model, or a teaching-domain large language model obtained by fine-tuning teaching-domain data. For example, the large language model can be obtained by fine-tuning a pre-trained general-domain large language model by applying various sample test data and the answer data of various sample test data. This embodiment of the invention does not specifically limit this.

[0042] After the prompts and test question data are input into a large language model, the model can solve the test question data according to the solution requirements specified in the prompts, thereby obtaining and outputting the solution data. Here, the solution data may include at least one of the following: solution approach, solution steps, and test question answer.

[0043] Understandably, because the prompts enhance the relevance and adaptability to both the characteristics of the test data and the users, the answer data output by the large language model based on the prompts is also more adapted to the test data's attributes and the users' cognitive attributes. As a result, the quality of the answer data is improved, and the answer data is also easier for users to understand and accept.

[0044] In the method provided in this embodiment of the invention, based on the test item attributes of the test item data and the cognitive attributes of the test item data towards users, the relevance and adaptability of the prompts corresponding to the test item data are enhanced when considering the characteristics of both the test item data and the users. By inputting the resulting prompts into a large-scale language model, test item attributes and user cognitive attributes that are more adapted to the test item data can be obtained. This improves the quality of test item answers, makes the answer data easier for users to understand and accept, and enhances the learning assistance effect.

[0045] Based on the above embodiments, step 120, determining the prompt corresponding to the test question data based on the test question attributes and the user-oriented cognitive attributes of the test question data, includes: From the candidate prompt templates, select the prompt template that matches the question attributes of the question data; The cognitive attributes of the test questions are filled into the prompt template to obtain the prompt corresponding to the test questions.

[0046] Specifically, to obtain more targeted prompts, a variety of candidate prompt templates can be pre-built. These diverse candidate prompt templates can be built separately for a variety of question attributes; that is, a candidate prompt template corresponding to each question attribute can be built separately.

[0047] Understandably, for each question attribute, the corresponding candidate prompt template can specifically describe the needs when answering question data under that question attribute. For example, for math multiple-choice questions, the corresponding candidate prompt template can specifically describe the needs when answering data-driven multiple-choice questions; for Chinese composition questions, the corresponding candidate prompt template can specifically describe the needs when answering Chinese composition questions.

[0048] Therefore, after obtaining the test data to be answered, a candidate prompt template corresponding to the test attribute of the test data can be selected from the pre-constructed candidate prompt templates. This template is then used as the prompt template matching the test attribute. For example, if the test data to be answered is a math multiple-choice question, the prompt template "You are the math teacher of [Grade X]. Explain the steps for multiple-choice questions: 1) Read the question and understand the knowledge points; 2) Analyze the differences between the options; 3) Use the elimination method; 4) Give the answer and reason; 5) Explain that other options are incorrect" can be selected.

[0049] Based on this, the cognitive attributes of the test data for users can be filled into a prompt template that matches the test attributes, thus obtaining the prompt corresponding to the test data. For example, if the cognitive attribute of the test data for users is grade 5, then "[Grade X]" in the prompt template in the above example can be filled with "Grade 5".

[0050] Understandably, by filling the cognitive attributes of the test data into the prompt template that matches the test attributes, the resulting prompts not only enhance the customization needs when answering test data, but also carry the cognitive attributes of the test data. Therefore, when large language models answer test data based on prompts, they can better adapt to the test attributes of the test data and the cognitive attributes of the users.

[0051] Based on any of the above embodiments, the prompt template includes a description of the solution steps for the question data under the question attribute.

[0052] Specifically, the prompt template that matches the question attribute of the question data to be answered may include text describing the solution steps for the question data under that question attribute.

[0053] Here, the solution step description text is natural language text that describes the steps taken to solve the test data under this test attribute. The solution step description text not only provides the steps taken to solve the test data under this test attribute, but also the order in which the steps are performed. By including the solution step description text in the prompt template, the large language model can be guided to execute the steps in the order described in the solution step description text, thereby achieving a more accurate solution to the test data that matches the test attributes, thus ensuring the quality of the solution data output by the large language model.

[0054] For example, for multiple-choice math questions, the prompt template could describe the solution steps as follows: "1) Read the question and understand the knowledge points; 2) Analyze the differences between the options; 3) Use the elimination method; 4) Give the answer and reason; 5) Explain why other options are incorrect." For fill-in-the-blank math questions, the prompt template could describe the solution steps as follows: "1) Analyze the question type; 2) Identify the core knowledge points; 3) Provide a solution strategy; 4) Write the answer in a standardized way; 5) Verify the correctness." For calculation math questions, the prompt template could describe the solution steps as follows: "1) List the known information and requirements; 2) Choose the formula or theorem; 3) Standardize the calculation process; 4) Check the result; 5) Summarize the method." For problem-solving math questions, the prompt template could describe the solution steps as follows: "1) Understand the question; 2) Analyze the conditions and the problem; 3) Develop a solution plan; 4) Write the complete process; 5) Check the answer; 6) Summarize the method." For proof math questions, the prompt template could describe the solution steps as follows: "1) Understand the proof conclusion; 2) Analyze the known information and the conclusion; 3) Find the proof strategy; 4) Write logically; 5) Check the rigor; 6) Summarize the techniques."

[0055] For example, for Chinese reading comprehension questions, the prompt template could describe the steps in the answer, such as "1) Overall understanding of the article 2) Close reading and answering questions 3) Analyzing key words and sentences 4) Grasping the main idea 5) Summarizing reading methods"; for classical Chinese reading comprehension questions, the prompt template could describe the steps in the answer, such as "1) Correcting pronunciation and characters 2) Understanding the meaning of the text 3) Analyzing the structure and content 4) Appreciating the thoughts and feelings 5) Summarizing learning methods"; for Chinese composition questions, the prompt template could describe the steps in the answer, such as "1) Analyzing the topic and establishing the theme 2) Selecting materials and conceiving the composition 3) Guiding the beginning and ending 4) Improving language expression 5) Teaching revision methods"; and for classical Chinese poetry appreciation questions, the prompt template could describe the steps in the answer, such as "1) Understanding the author's background 2) Correcting pronunciation and understanding word meanings 3) Understanding the main idea of ​​the poem 4) Appreciating the artistic conception and emotions 5) Appreciating the language characteristics 6) Summarizing the expressive techniques".

[0056] For example, for English cloze tests, the prompt template could describe the steps in the solution process as follows: "1) Read through and grasp the theme; 2) Analyze the logic of each blank; 3) Choose the appropriate answer; 4) Check the rationality; 5) Summarize the problem-solving strategies." For English reading comprehension questions, the prompt template could describe the steps in the solution process as follows: "1) Quickly skim to understand the main idea; 2) Carefully read and understand the details; 3) Analyze the language features; 4) Find the basis for answering the question; 5) Summarize the methods and techniques." For English essay questions, the prompt template could describe the steps in the solution process as follows: "1) Examine the question and analyze the requirements; 2) conceive the essay structure; 3) Use vocabulary and sentence patterns; 4) Check the grammar for correctness; 5) Refine and polish the expression."

[0057] In the method provided in this embodiment of the invention, by adding a description of the solution steps for the test data under the test data attributes to the prompt template, the large language model can be guided to execute the steps in the order described in the description of the solution steps, so as to achieve a test answer that is more in line with the test data attributes, thereby ensuring the quality of the solution data output by the large language model.

[0058] Based on any of the above embodiments, the solution step description text includes a knowledge point extraction step description text, which describes the steps of extracting the knowledge points required to answer the test data from knowledge points that conform to the cognitive attributes.

[0059] Specifically, knowledge points can be knowledge units such as theories, formulas, axioms, and theorems. To ensure that the solution data output by the large language model matches the user's cognitive level and is thus understandable and acceptable to the user, a knowledge point extraction step can be set in the solution step description text. This is specifically manifested in the inclusion of a knowledge point extraction step description text within the solution step description text.

[0060] Here, the description text of the knowledge point extraction steps is specifically the natural language text that describes the steps of knowledge point extraction. For example, in the description text of the solution steps for a math multiple-choice question, the description text of the knowledge point extraction steps could be "read the question and understand the knowledge points"; in the description text of the solution steps for a math calculation question, the description text of the knowledge point extraction steps could be "select formulas and theorems".

[0061] By including knowledge point extraction step description text in the solution step description text, a large language model can be guided to extract the knowledge points required to answer the test questions from the knowledge points that match the user's cognitive attributes filled in the prompts. This ensures that the knowledge points applied by the large language model when answering the test questions match the user's cognitive attributes, and thus ensures that the solution data output by the large language model can be understood and accepted by the user.

[0062] Based on any of the above embodiments, the method further includes: Based on the user's registration information and / or the user's historical learning records, the cognitive attributes of the user's test data are determined.

[0063] Specifically, before determining the prompts corresponding to the test question data, the cognitive attributes of the test question data for the user can be determined first. That is, the cognitive attributes can be determined based on the user's registration information, or based on the user's historical learning records, or a combination of the user's registration information and historical learning records.

[0064] Registration information refers to the personal information a user enters into their device when registering an account. This information may include the user's date of birth, educational background, age, and other details that can be used to infer cognitive attributes. Therefore, based on the test data and the user's registration information, the user's academic year can be inferred, thus revealing their cognitive attributes.

[0065] Historical learning records refer to the information recorded by a user through device learning in the past. This may include test data entered by the user in the past, interaction data between the user and the device regarding learning-related content, and test scores entered by the user in the past—information that can be used to infer cognitive attributes. Therefore, based on test data and a user's historical learning records, the academic year the user is in can be inferred, thereby obtaining the user's cognitive attributes. For example, the user's cognitive attributes can be determined based on the academic year to which the knowledge points tested in the user's historically entered test data belong; alternatively, the user's cognitive data can be identified based on historical interaction data with the device regarding learning-related content. This embodiment of the invention does not specifically limit this approach.

[0066] Based on any of the above embodiments, the test item attributes include subject and / or question type.

[0067] Specifically, when determining the prompts corresponding to test question data, the test question attributes that can be referenced can include subject and / or question type. It is understandable that different subjects may require different knowledge points and problem-solving approaches, and even within the same subject, different question types may require different problem-solving approaches. Therefore, the prompts corresponding to the test question data can be determined by considering the subject and / or question type of the test question data and the cognitive attributes of the test question data intended for the user.

[0068] For example, when answering reading comprehension questions in Chinese, the common problem-solving approach is to perceive → read carefully → analyze → grasp → summarize. However, when answering reading comprehension questions in English, the common approach is to browse → read carefully → analyze → search → summarize. Therefore, different prompts can be used for test data in different subjects.

[0069] For example, even within the same subject of mathematics, the common problem-solving approach for calculation problems is to set up the equation → use the formula → calculate → check → summarize. However, the common approach for proof problems in data science is to understand → analyze → think through the problem → write it down → check → summarize. In other words, different types of questions can correspond to different prompts.

[0070] In some embodiments, corresponding candidate prompt templates can be pre-set for different subjects and question types. The specific format could be "You are a teacher of [Grade X] [Subject]. Explaining the steps for [Question Type]:...". It is understood that "[Grade X]" is used to fill in the cognitive attributes of the test data for the user; "[Subject]" is used to identify the subject corresponding to the candidate prompt template, such as mathematics, Chinese, English, physics, history, etc.; and "[Question Type]" is used to identify the question type corresponding to the candidate prompt template, such as multiple choice, fill-in-the-blank, calculation, problem-solving, proof, reading comprehension, classical Chinese reading, essay, classical poetry appreciation, cloze test, etc.

[0071] For example, for multiple-choice math questions, the typical problem-solving approach is understanding → analysis → elimination → verification → summarization. A template for candidate prompts based on this approach would be: "You are the math teacher for [Grade X]. Here are the steps to explain multiple-choice questions: 1) Read the question and understand the relevant concepts; 2) Analyze the differences between the options; 3) Use the elimination method; 4) Provide the answer and reasons; 5) Explain why the other options are incorrect."

[0072] For math fill-in-the-blank questions, the typical problem-solving approach is understanding → analysis → elimination → verification → summarization. A template for candidate prompts based on this approach would be: "You are a math teacher for [Grade X]. The steps for explaining fill-in-the-blank questions are: 1) Analyze the question type; 2) Identify the core knowledge points; 3) Provide a problem-solving approach; 4) Write the answer correctly; 5) Verify its correctness."

[0073] For math calculation problems, the typical problem-solving approach is: list the equation → use the formula → calculate → check → summarize. A possible prompt template would be: "You are a math teacher for [Grade X]. Here are the steps to explain calculation problems: 1) List the known information and requirements; 2) Choose the appropriate formula or theorem; 3) Properly demonstrate the calculation process; 4) Check the results; 5) Summarize the method."

[0074] For math problem-solving questions, the typical approach is understanding → analysis → planning → writing → checking → summarizing. A possible prompt template would be: "You are the math teacher for [Grade X]. Here are the steps to explain problem-solving: 1) Understand the problem; 2) Analyze the conditions and the question; 3) Develop a problem-solving plan; 4) Write out the complete solution; 5) Check the answer; 6) Summarize the method."

[0075] For mathematical proof problems, the typical problem-solving approach is: understanding → analysis → thought process → writing → checking → summarizing. A possible prompt template is: "You are a math teacher for [Grade X]. Here are the steps to explain a proof problem: 1) Understand the conclusion; 2) Analyze the given information and the conclusion; 3) Find the proof's thought process; 4) Write the proof logically; 5) Check for rigor; 6) Summarize the techniques."

[0076] For Chinese reading comprehension questions, the typical problem-solving approach is: perception → close reading → analysis → understanding → summarizing. A possible prompt template is: "You are the Chinese teacher for [Grade X]. Here are the steps to explain reading comprehension questions: 1) Gain an overall understanding of the article; 2) Read carefully and answer the questions; 3) Analyze key words and sentences; 4) Grasp the main theme; 5) Summarize reading methods."

[0077] For classical Chinese reading comprehension questions, the typical problem-solving approach is: correct pronunciation → clarify understanding → analyze → appreciate → summarize. A possible prompt template would be: "You are a [Grade X] Chinese teacher. The steps for explaining classical Chinese reading comprehension are: 1) Correct pronunciation and characters; 2) Clarify the meaning of the text; 3) Analyze the structure and content; 4) Appreciate the thoughts and feelings expressed; 5) Summarize learning methods."

[0078] For Chinese composition questions, the typical problem-solving approach is: analyze the question → select materials → structure → express → revise. A possible prompt template would be: "You are the Chinese teacher for [Grade X]. Here are the steps to explain the composition question: 1) Analyze the question and establish the theme; 2) Select materials and develop the idea; 3) Guide the beginning and ending; 4) Improve language expression; 5) Teach revision methods."

[0079] For questions on appreciating classical Chinese poetry, the typical approach is: background → pronunciation correction → comprehension → understanding → appreciation → summary. A possible prompt template would be: "You are a [Grade X] Chinese teacher. Here are the steps to explain appreciating classical Chinese poetry: 1) Understand the author's background; 2) Correct pronunciation and understand word meanings; 3) Comprehend the main idea of ​​the poem; 4) Appreciate the imagery and emotions; 5) Appreciate the language features; 6) Summarize the expressive techniques."

[0080] For English cloze tests, the typical problem-solving approach is: grasp the main idea → analyze → select the correct answer → check the answer → develop a strategy. A sample prompt template would be: "You are the English teacher for [Grade X]. Here are the steps to answer a cloze test: 1) Read through the text to grasp the main idea; 2) Analyze the logic of each blank; 3) Select the appropriate answer; 4) Check the answer's validity; 5) Summarize the problem-solving strategies."

[0081] For English reading comprehension questions, the typical problem-solving approach is: skim → read carefully → analyze → find the answer → summarize. A possible prompt template would be: "You are the English teacher for [Grade X]. Here are the steps to explain reading comprehension questions: 1) Quickly skim to understand the main idea; 2) Carefully read and understand the details; 3) Analyze the language features; 4) Find the supporting evidence to answer the questions; 5) Summarize the methods and techniques."

[0082] For English essay questions, the typical problem-solving approach is: analyze the question → brainstorm → apply the ideas → check → refine. A possible prompt template is: "You are the English teacher for [Grade X]. Here are the steps to explain the essay question: 1) Analyze the question requirements; 2) Brainstorm the essay structure; 3) Use vocabulary and sentence structures; 4) Check for grammatical correctness; 5) Refine and polish the expression."

[0083] Based on this, after obtaining the test question data entered by the user, the subject and question type of the test question data can be determined, as well as the user's cognitive attributes.

[0084] Therefore, based on the subject and question type of the test data, a prompt template matching the subject and question type of the test data can be selected from candidate prompt templates of various subject and question type combinations, and the user's cognitive attributes can be filled into the prompt template, thereby obtaining the prompt corresponding to the test data to be answered.

[0085] Then, the test question data to be answered, along with the corresponding prompts, can be input into the large language model to obtain the answer data output by the large language model, and then the answer data can be displayed.

[0086] In the method provided in this embodiment of the invention, prompts are determined by the subject, question type, and user cognitive attributes of the test data. A large language model is then used to generate a problem-solving process that conforms to the user's cognitive attributes based on the prompts, so that the automated test answer can better meet the user's actual needs and optimize the experience of automated question searching and answering.

[0087] The test question answering device provided by the present invention is described below. The test question answering device described below can be referred to in correspondence with the test question answering method described above.

[0088] Figure 2 This is a schematic diagram of the test question answering device provided by the present invention, as shown below. Figure 2 As shown, the device includes: Unit 210 is used to obtain the test data to be answered; The prompt construction unit 220 is used to determine the prompt corresponding to the test question data based on the test question attributes of the test question data and the user-oriented cognitive attributes of the test question data; The solution unit 230 is used to input the test question data and the prompts into a large language model to obtain the solution data of the test question data output by the large language model.

[0089] In the apparatus provided in this embodiment of the invention, based on the test item attributes of the test item data and the cognitive attributes of the test item data towards the user, the relevance and adaptability of the prompts corresponding to the test item data are enhanced when considering the characteristics of both the test item data and the user. By inputting the resulting prompts into a large-scale language model, test item attributes and user cognitive attributes that are more adapted to the test item data can be obtained. This improves the quality of test item answers, makes the answer data easier for users to understand and accept, and enhances the learning assistance effect.

[0090] Based on the above embodiments, the prompt construction unit is specifically used for: From the candidate prompt templates, select the prompt template that matches the question attributes of the question data; The cognitive attributes of the test questions are filled into the prompt template to obtain the prompt corresponding to the test questions.

[0091] Based on any of the above embodiments, the prompt template includes a description of the solution steps for the question data under the question attribute.

[0092] Based on any of the above embodiments, the solution step description text includes a knowledge point extraction step description text, which describes the steps of extracting the knowledge points required to answer the test data from knowledge points that conform to the cognitive attributes.

[0093] Based on any of the above embodiments, the device further includes a cognitive determination unit, used for: Based on the user's registration information and / or the user's historical learning records, the cognitive attributes of the user's test data are determined.

[0094] Based on any of the above embodiments, the test item attributes include subject and / or question type.

[0095] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 310, a communications interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communications interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute a test question solution method, which includes: Obtain the data of unanswered test questions; Based on the test question attributes of the test question data and the cognitive attributes of the test question data towards users, determine the prompts corresponding to the test question data; The test question data and the prompts are input into a large language model to obtain the answer data of the test question data output by the large language model.

[0096] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute the test question answering methods provided by the above methods, the method comprising: Obtain the data of unanswered test questions; Based on the test question attributes of the test question data and the cognitive attributes of the test question data towards users, determine the prompts corresponding to the test question data; The test question data and the prompts are input into a large language model to obtain the answer data of the test question data output by the large language model.

[0098] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the test question-solving methods provided by the methods described above, the method comprising: Obtain the data of unanswered test questions; Based on the test question attributes of the test question data and the cognitive attributes of the test question data towards users, determine the prompts corresponding to the test question data; The test question data and the prompts are input into a large language model to obtain the answer data of the test question data output by the large language model.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of software products. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A test question answering method characterized by comprising: include: Obtain the data of unanswered test questions; Based on the test question attributes of the test question data and the cognitive attributes of the test question data towards users, determine the prompts corresponding to the test question data; The test question data and the prompts are input into a large language model to obtain the answer data of the test question data output by the large language model.

2. The test question answering method according to claim 1, characterized in that, The process of determining the prompts corresponding to the test questions based on the test question attributes and the user-oriented cognitive attributes of the test question data includes: From the candidate prompt templates, select the prompt template that matches the question attributes of the question data; The cognitive attributes of the test questions are filled into the prompt template to obtain the prompt corresponding to the test questions.

3. The test question answering method according to claim 2, characterized in that, The prompt template includes a text describing the solution steps for the question data under the specified question attribute.

4. The test question answering method according to claim 3, characterized in that, The solution step description text includes a knowledge point extraction step description text, which describes the steps of extracting the knowledge points needed to answer the test questions from knowledge points that conform to the cognitive attributes.

5. The test question answering method according to any one of claims 1 to 4, characterized in that, Also includes: Based on the user's registration information and / or the user's historical learning records, the cognitive attributes of the user's test data are determined.

6. The test question answering method according to any one of claims 1 to 4, characterized in that, The test item attributes include subject and / or question type.

7. A test question answering device, characterized in that, include: The acquisition unit is used to acquire the test question data to be answered; The prompt construction unit is used to determine the prompt corresponding to the test question data based on the test question attributes and the user-oriented cognitive attributes of the test question data. The solution unit is used to input the test question data and the prompts into a large language model to obtain the solution data of the test question data output by the large language model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the test question answering method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the test question answering method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the test question answering method as described in any one of claims 1 to 6.

Images