Title processing method and apparatus

Abstract

This application provides a question processing method and apparatus. The question processing method includes: acquiring question data to be processed, wherein the question data includes at least one question to be processed; inputting the question data and prompt text into a question processing model to obtain at least one target question output by the question processing model, wherein the question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question, and to determine the hierarchical relationship between the questions based on the question number and question type corresponding to each question, thereby obtaining at least one target question. According to the question processing method provided by this application, the complex logic inherent in the questions can be accurately identified and transformed.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a problem-solving method. This application also relates to a problem-solving apparatus, a computing device, a computer-readable storage medium, and a computer program product. Background Technology

[0002] With the continuous development of computer technology, in the field of intelligent education, a massive amount of academic questions need to be transformed from original paper documents, scanned images, or unstructured web page texts into high-quality structured database resources.

[0003] During this process, due to the high heterogeneity of question types and formats, the conversion process usually only allows for preliminary character recognition of the questions, making it difficult to accurately identify and convert the complex logic within the questions. Summary of the Invention

[0004] In view of this, embodiments of this application provide a problem processing method. This application also relates to a problem processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, to solve the aforementioned problems existing in the prior art.

[0005] According to a first aspect of the embodiments of this application, a question processing method is provided, including: Obtain the data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed; The question data to be processed and the prompt text are input into the question processing model to obtain at least one target question output by the question processing model. The question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question. Based on the question number and question type corresponding to each question, the hierarchical relationship between each question is determined to obtain at least one target question.

[0006] According to a second aspect of the embodiments of this application, a question processing apparatus is provided, comprising: The acquisition unit is configured to acquire data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed; The processing unit is configured to input the question data to be processed and the prompt text into the question processing model to obtain at least one target question output by the question processing model. The question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question. Based on the question number and question type corresponding to each question, the hierarchical relationship between each question is determined to obtain at least one target question.

[0007] According to a third aspect of the embodiments of this application, a computing device is provided, comprising: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions, which, when executed by the processor, implement the steps of the above-described problem-solving method.

[0008] According to a fourth aspect of the embodiments of this application, a computer-readable storage medium is provided that stores a computer program / instructions, which, when executed by a processor, implement the steps of the above-described problem-solving method.

[0009] According to a fifth aspect of the embodiments of this application, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described problem processing method.

[0010] According to the question processing method provided in this application, question data containing at least one question to be processed is obtained. In the question processing model, the question number and question type corresponding to each question data item are parsed based on the prompt text. This process, extracting the question number and question type from the question data based on the prompt text, achieves structured processing of the question data, facilitating the determination of the hierarchical relationship between different questions in the question data based on the question number and question type. This achieves the goal of accurately identifying the inherent logic. Attached Figure Description

[0011] Figure 1 A flowchart of a question processing method according to an embodiment of this application is shown; Figure 2 A flowchart illustrating a problem-solving method for problem processing according to an embodiment of this application is shown. Figure 3 This invention provides a schematic diagram of the structure of a question processing device according to an embodiment of the present application. Figure 4 A structural block diagram of a computing device according to an embodiment of this application is shown. Detailed Implementation

[0012] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0013] The terminology used in one or more embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this application. The singular forms “a,” “the,” and “the” used in one or more embodiments of this application and in the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” used in one or more embodiments of this application refers to and includes any or all possible combinations of one or more associated listed items.

[0014] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this application, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0015] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0016] The question processing method involved in this application is applied to scenarios where unstructured question data is structured according to the logical relationships between questions.

[0017] With the continuous development of computer technology, in the field of intelligent education, a massive amount of academic questions need to be transformed from raw paper documents, scanned images, or unstructured web page text into high-quality structured database resources. In this process, the question data may consist of a mixture of handwritten formulas, printed text, and geometric diagrams. Furthermore, complex situations exist such as "multiple questions for one problem," "multiple solutions for a single blank," and "a mixture of solution processes and final values." Therefore, educational data is typically highly heterogeneous.

[0018] Given the above, educational data is typically extracted using rule-based matching and pattern recognition. However, this approach relies on manually pre-defined keyword libraries and regular expressions. For example, common answer prompts are used to locate answer ranges, and specific matching patterns are used to attempt to extract numerical values ​​or option labels. When faced with questions that have varied natural language descriptions, or when answers are hidden at the end of complex derivations, rule-based matching often fails to cover all language expressions, leading to extraction failures. Furthermore, a step-by-step processing approach combining OCR with traditional NLP pipelines can be used to achieve structured extraction and recognition of educational data. This approach first uses a visual model to extract text from images, then uses a lightweight text classification model to determine the question type, and finally uses named entity recognition technology to attempt to extract core numerical values ​​from the text stream. However, while this pipeline architecture offers flexibility, it suffers from a significant error accumulation effect.

[0019] In view of this, this application provides a question processing method, which inputs the question data to be processed into a question processing model. In the question processing model, through multimodal perception of the question data, logical decomposition and formatted rewriting based on the question number are performed to obtain a structured target question, thereby achieving accurate identification and transformation of the complex logic inherent in the question. This application also relates to a question processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, which will be described in detail in the following embodiments.

[0020] Figure 1 A flowchart of a question processing method according to an embodiment of this application is shown, specifically including the following steps 102-104: Step 102: Obtain the data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed.

[0021] The question data to be processed can be understood as the original question information. It should be understood that the question data to be processed involved in this application can exist in various forms. For example, it can be an image containing questions (e.g., a scanned exam paper), a piece of text containing questions (e.g., questions on a webpage), or data that combines images and text.

[0022] A question to be processed can be understood as an independent question unit contained within the question data. A dataset can contain one or more such units.

[0023] In one specific embodiment provided in this application, data of questions to be processed, containing at least one question to be processed, is obtained. It should be understood that the data of questions to be processed can be question information in any form. For example, the data of questions to be processed can be an image containing only one question, or it can be an image containing multiple questions.

[0024] Furthermore, this application does not limit the format of the data to be processed. For example, it can be in image format or text format, etc.

[0025] Step 104: Input the question data to be processed and the prompt text into the question processing model to obtain at least one target question output by the question processing model.

[0026] In this process, the question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question. Based on the question number and question type corresponding to each question, the hierarchical relationship between the questions is determined to obtain at least one target question.

[0027] The prompt text can be understood as a set of natural language instructions that guide the question processing model on how to work. The prompt text instructs the model on what tasks it needs to complete and the format and rules that its output should follow.

[0028] In one example, the prompt text could include instructions such as "Please identify all questions", "Determine the type of each question (single choice, multiple choice, fill-in-the-blank, solution)", and "Find the parent-child relationship between questions".

[0029] The question processing model can be understood as a multimodal large language model (MLLM) with comprehension capabilities. This model can understand images and text, and can perform in-depth analysis and processing of the input question data based on the guidance of the prompt text.

[0030] Target questions can be understood as questions with structured information output after being processed by the question processing model. Compared to the original questions to be processed, target questions have been clearly identified, classified, and logical connections have been established between different target questions.

[0031] The question number can be understood as an identifier that comes with the question and is used to distinguish different questions. For example, the question number can be: "1", "(1)", "①", "Question 2", etc.

[0032] Question types can be understood as a classification of the nature of questions. For example, question types can be: multiple choice, true / false, fill-in-the-blank, problem-solving, proof, etc.

[0033] Hierarchical relationships can be understood as the logical structure between questions. For example, a main question (parent question) may contain several sub-questions (child questions), or several independent questions.

[0034] In one specific embodiment provided in this application, a question processing model is used to accurately identify each independent question from the question data to be processed. Then, the question number and question type corresponding to each independent question are identified. Based on this, the hierarchical relationship between the questions is determined according to the question number and question type corresponding to each independent question; for example, the belonging relationship between questions. This results in at least one target question with a clear structure.

[0035] To facilitate understanding, the problem-solving methods are explained in the following manner.

[0036] In one example, a question processing model is built to facilitate the structured processing of questions based on the question processing model.

[0037] In one specific embodiment provided in this application, the question processing model can be pre-built in the following manner: Obtain initial sample question data and input it into the initial question processing model; In the initial question processing model, the question storage format corresponding to the initial sample question data is determined, and the sample question data is determined based on the question storage format; based on the prompt text, the sample question data is parsed to obtain at least one question, as well as the question number and question type corresponding to each question; based on the question number and question type corresponding to each question, the hierarchical relationship between each question is determined to obtain at least one predicted sample question; The initial question processing model is trained based on the predicted sample questions until the training stopping condition of the question processing model is met, thus obtaining the question processing model.

[0038] The initial sample question data can be understood as the raw data used to train the model. This data consists of known question information with correct answers or standard structures, and can be in the form of images, text, etc.

[0039] The initial question processing model can be understood as a model that has not yet been fully trained and has little or no question parsing ability.

[0040] The sample question data can be understood as a text stream that has undergone preliminary format standardization and is directly input into the initial question processing model for parsing.

[0041] The prompt text can be understood as instructions given during the training process to guide the initial question processing model on how to parse the questions.

[0042] Predicted sample questions can be understood as the structured question results output by the initial question processing model after it autonomously "thinks" based on the input sample data and prompt text.

[0043] Training termination criteria can be understood as metrics used to determine whether the model has completed its learning process. For example, the model's accuracy on the validation set no longer improves, or it has reached the preset number of training epochs. Once this condition is met, the training process ends.

[0044] Specifically, the process involves acquiring initial sample question data and, based on the initial question processing model, first determining whether the input sample is an image or text, then preprocessing it to unify it into processable "sample question data." The initial question processing model then parses the processed data according to the instructions in the prompt text, attempting to identify each question, its number, and its type. Based on this, the initial question processing model constructs a hierarchical relationship between the questions according to the identified numbers and types. Finally, for this sample, the question processing model outputs predicted sample questions.

[0045] Building upon the above, the predicted sample questions output by the initial question processing model are compared with the actual results inherent in the sample data (e.g., manually labeled question numbers, types, and hierarchical relationships), and the difference between the two (e.g., the loss function value) is calculated. A larger difference indicates a worse model performance. Based on the calculated difference, billions or even hundreds of billions of parameters within the initial question processing model are adjusted using the backpropagation algorithm, so that the difference between the model's output and the actual results becomes smaller when it encounters similar data again. Steps 1-3 are repeated, using tens of thousands of samples to repeatedly train the model and adjust its parameters. When the model's performance is good enough to meet the preset "training stopping condition," training ends. The model at this point is the final usable question processing model.

[0046] Furthermore, after obtaining the trained question processing model based on the above, this application uses the following method to obtain at least one target question based on the question processing model.

[0047] In one specific embodiment provided in this application, at least one target question is obtained from the output of the question processing model, including S1042-S104: S1042. Determine the question storage format corresponding to the question data to be processed, and determine the target question data based on the question storage format.

[0048] The question storage format can be understood as the format in which the question data to be processed exists. For example, the question storage format can be an image format or a plain text format.

[0049] The target question data can be understood as the data that, after preliminary processing, is used to input into the question processing model. It should be understood that the target question data in this application is in a uniform text format.

[0050] In one specific embodiment provided in this application, it is first determined whether the question data to be processed is stored in the format of an image or text. Then, appropriate conversion or cleaning is performed according to its format to ensure that the subsequent question processing model can handle it.

[0051] In one example, if the questions in the data to be processed are stored in image format, the text can be extracted by calling the OCR function to form text.

[0052] Specifically, this application explains the method for determining the target topic data in conjunction with the following methods.

[0053] In one specific embodiment provided in this application, determining the target question data based on the question storage format includes: If the question data to be processed is stored in image format, the image format is converted into text format to obtain the question data to be processed, and the target question data is obtained based on the question data to be processed and the set rules. If the question data to be processed is stored in text format, the text format is obtained based on the set rules to obtain the question data to be processed, and the target question data is obtained based on the question data to be processed and the set rules.

[0054] The rules set here can be understood as rules used to clean and preprocess text data. For example, removing meaningless line breaks and spaces, or identifying and retaining special symbols in formulas.

[0055] In one specific embodiment provided in this application, for the question data to be processed stored in image format, the image to text conversion (such as OCR) is first performed, and then the obtained text is optimized and cleaned according to the set rules, such as correcting recognition errors (such as correcting "O" to "0"), to obtain text data, which is then used as the target question data.

[0056] Furthermore, if the input question data is already in text format, the set rules are directly applied to clean and optimize it, removing format noise to obtain text data, which is then used as the target question data.

[0057] According to a specific implementation method provided in this application, the input format is processed separately. For the question data to be processed input in image format, after OCR recognition, error correction is performed according to the set rules to achieve secondary optimization. This avoids the situation where OCR recognition errors directly lead to subsequent logical errors. By correcting errors at the semantic level, the accuracy and robustness of the processing chain are improved.

[0058] Furthermore, based on the question data to be processed and the set rules, the target question data is obtained, including: Obtain the semantic information corresponding to the question data to be processed; Based on the semantic information, the data context corresponding to the question data to be processed is determined; Based on the data context, the question data to be processed is modified to obtain the target question data.

[0059] In this context, data context can be understood as the specific environment in which the data to be processed exists, used to characterize the subject type of the data. For example, in the data to be processed corresponding to the subject type of mathematics, the letter "e" may be a variable or it may represent a natural constant. In this case, its specific meaning is determined based on the data context.

[0060] In one specific embodiment provided in this application, a question processing model is used to understand the semantic information corresponding to the question data to be processed, so as to determine the context of the question data. Further, based on the obtained context of the question data, the precise context of the question data is determined. For example, the subject corresponding to the question data is determined, thereby determining whether it is in the context of exponential operations or algebraic evaluation, etc. After obtaining the context of the question data, the question data is corrected using contextual knowledge. For example, in the context of exponential operations, the vaguely identified "e" is corrected to an accurate "e," etc. The corrected target question data is then obtained.

[0061] According to a specific implementation method provided in this application, the semantic understanding capability of the question processing model is utilized to first determine the context, and then the context is used to correct recognition errors, thereby enhancing the ability to process fuzzy and incomplete data.

[0062] S1044. Based on the prompt text, parse the target question data to obtain at least one question, as well as the question number and question type corresponding to each question.

[0063] In one specific embodiment provided in this application, the question processing model parses the target question data with uniform format according to the guidance of the prompt text, segments it into at least one independent question, and labels each independent question with its corresponding question number and question type.

[0064] Specifically, this application explains the method of annotating each target title in the following way.

[0065] In one specific embodiment provided in this application, based on the prompt text, the target question data is parsed to obtain at least one question, and the question number and question type corresponding to each question, including: Based on the prompt text, the target question data is parsed to obtain at least one question, the question number corresponding to each question, and the question semantics corresponding to each question are determined; Based on the semantics of each question, determine the instruction verbs corresponding to each question; Based on the instruction verbs corresponding to each question, determine the question type for each question.

[0066] The semantics of a question can be understood as the meaning expressed in the question's content. For example, does the question require "selection," "judgment," or "calculation"?

[0067] Instructional verbs can be understood as words in the question that explicitly indicate the process of answering the question, such as "choose," "judge," "calculate," "prove," "draw," etc. The type of question can be determined by the instructional verbs.

[0068] In one specific embodiment provided in this application, after identifying the question and question number, the question processing model further understands the semantics of each question. It then extracts instruction verbs from the understood semantics. For example, it extracts "select" from "Please select the correct answer". Finally, it maps the extracted instruction verbs to preset question types. For example, it maps "select" to "multiple choice question", "judgment" to "true / false question", and "calculate" or "solve" to "answer question".

[0069] According to a specific implementation method provided in this application, by understanding the "question semantics" and extracting the "instructional verbs," the question processing model can accurately classify the question based on the instruction verbs even if the question description changes. This improves the accuracy of question type judgment.

[0070] S1046. Based on the question number and question type corresponding to each question, determine the hierarchical relationship between the questions to obtain at least one target question.

[0071] In one specific embodiment provided in this application, the question processing model infers and determines the hierarchical relationship between each question, such as parent-child or sibling relationships, based on the question number (e.g., "(1)" below "I,") and question type (e.g., the type of a major question may be "compound question") obtained above, and finally obtains the structured target questions.

[0072] In one specific embodiment provided in this application, based on the question number and question type corresponding to each question, the hierarchical relationship between the questions is determined to obtain at least one target question, including: Based on the question type, identify at least one question corresponding to each question type; Based on the question number, the hierarchical relationship between the questions is determined, wherein the hierarchical relationship includes the hierarchical relationship between questions of the same question type, and the hierarchical relationship between questions corresponding to different question types; Sort at least one question of the same question type based on the hierarchical relationship between the questions of the same question type to obtain at least one target question of the same question type. The questions corresponding to different question types are sorted according to the hierarchical relationship between the questions of different question types to obtain at least one target question corresponding to each question type.

[0073] In one example, the hierarchical relationship between questions of the same question type can be that within a major question of the "compound question" type, the order and inclusion relationship between several sub-questions of the "fill-in-the-blank" type can be used.

[0074] In another example, the hierarchical relationship between questions of different question types could be that a "multiple choice" question and a "problem-solving" question are parallel and independent. Alternatively, it could be an inclusion relationship between a "compound question" (parent question, type compound) and its several "fill-in-the-blank" sub-questions (type fill-in-the-blank).

[0075] In one specific embodiment provided in this application, the questions are first grouped by type. For example, all "multiple choice questions" are placed in one group, and all "fill-in-the-blank questions" are placed in another group. Based on the grouping, the logical structure between all questions is further clarified using the question numbers. The questions are then sorted within the same group, or logical connections are established between different groups. The grouped questions are sorted according to their internal numbering order (e.g., (1), (2), (3)) to ensure that the order within each type of question is correct and coherent. Following the overall logic of the entire question set (e.g., the question numbering order 1, 2, 3... of the exam paper), all different types of questions (including parent and child questions) are integrated to form a complete and correctly structured question list.

[0076] According to a specific implementation method provided in this application, by format judgment and conversion, it is ensured that various complex input sources (images, text, etc.) can be processed.

[0077] In one specific embodiment provided in this application, the at least one target topic obtained above in this application can be any target topic stored in a topic list according to the hierarchical relationship between the topics.

[0078] Furthermore, in a specific embodiment provided in this application, the method for obtaining each target title further includes: Identify the target question to be processed, and obtain the target question stem and initial reference answer corresponding to the target question to be processed, wherein the target question to be processed is any one of the target questions; Determine the question stem attribute information corresponding to the target question stem; Based on the aforementioned question stem attribute information, a predicted reference answer is determined; Based on the predicted reference answer, the initial reference answer is adjusted to obtain the target reference answer.

[0079] Among them, the target questions to be processed can be understood as questions that have been processed by the aforementioned steps, but have not yet undergone answer verification and standardization.

[0080] The target question stem can be understood as the question portion of the question, which does not include the answer.

[0081] The initial reference answer can be understood as the answer that was originally provided with the question.

[0082] The information in the question stem can be understood as the requirements for the answer that can be inferred from the question stem. For example, from "find the slope k and intercept b of the line", we can infer that the question requires two answers.

[0083] In one specific embodiment provided in this application, after the structured parsing of the questions is completed, a single question is processed. First, the question stem and its original answer are obtained. The question processing model rereads the question stem to analyze how many answers the question theoretically requires. Based on the analysis, a prediction of the theoretical number or form of answers is generated, resulting in a predicted reference answer. For example, the prediction result is "this question should have 2 answers." The original initial reference answer is verified and adjusted using the theoretical prediction value. If the original initial reference answer only provides one value, but the theoretical prediction requires two, it can be determined that the original initial reference answer may be missing, and it is marked or corrected. A verified target reference answer that meets the theoretical expectation is obtained.

[0084] According to a specific implementation method provided in this application, by introducing question stem attribute information and predicted reference answers, it is possible to automatically detect problems such as missing answers (e.g., only one blank is written in a fill-in-the-blank question) and multiple answers in the original data, thus solving the problem of not being able to associate with the corresponding sub-questions and missing answers.

[0085] Furthermore, regarding obtaining each target question, the method also includes: Determine the initial target question, wherein the initial target question is any one of the target questions; Obtain the question representation format corresponding to the initial target question; Based on the preset format, the question representation format corresponding to the initial target question is adjusted to obtain the initial target question in the preset format.

[0086] The question representation format can be understood as the current presentation format of the question content and the answer. For example, the answer might be a piece of natural language text such as "Therefore, the answer is 5," and a true / false question might be "..." " or "T".

[0087] A preset format can be understood as a pre-defined, standardized output format. In this application, it could refer to, for example, LaTeX format and structured JSON format. For instance, all mathematical symbols must be enclosed in $, and true / false symbols must be enclosed in $. "It should be uniformly converted to \checkmark."

[0088] In one specific embodiment provided in this application, a question that has already undergone structured parsing is selected as the processing object. The current presentation format of this question is examined. The content and answer of the question are converted into a preset standard format. For example, "So the answer is 5" is converted to "5" (with a $ symbol), "correct" is converted to \checkmark, and then all information is packaged into a well-structured JSON object.

[0089] According to a specific implementation provided in this application, answer expressions (such as "correct", "T", "√") are standardized into machine-readable, internationally standard LaTeX notation (\checkmark) through formatting, and all explanatory text is stripped away, retaining only the core numerical values. All information is then encapsulated in structured JSON. This allows the output data to be directly read by a database and seamlessly accessed by downstream systems.

[0090] The following is in conjunction with the appendix Figure 2 Taking the application of the question processing method provided in this application as an example, the question processing method will be further explained. Figure 2 This application provides a flowchart illustrating a problem-solving method according to an embodiment of the present application, which specifically includes the following steps: Step 202: Obtain the data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed.

[0091] In one specific embodiment provided in this application, image or text data is used as the question data to be processed, and at least one question is extracted from the image or text data.

[0092] Step 204: Determine the target question data based on the question data to be processed.

[0093] In this application, for example, based on a question processing model, at least one target question data is determined according to the question data to be processed.

[0094] In one example, when the question data to be processed is an image, the image and the prompt text are input into the question processing model. The question processing model recognizes the content in the image and obtains the target question data in text format.

[0095] In another example, when the question data to be processed is text data, the text data and the prompt text are input into the question processing model. In the question processing model, the text data is subjected to content recognition to obtain the target question data in text format.

[0096] After obtaining the question data to be processed, the semantic context capability of the question processing model is used to determine the data context of the question data to be processed. Based on the data context, the question data to be processed is adjusted to obtain the target question data.

[0097] Step 206: Extract from the target question data according to the set rules to obtain at least one target question.

[0098] In one specific embodiment provided in this application, in the question processing model, target question data is extracted based on the prompt text to obtain at least one target question.

[0099] In one example, the question numbers in the target question data are scanned to obtain at least one initial target question. Based on the question numbering strategy, the hierarchical relationship between the questions in the target question data is determined to obtain at least one target question. It is important to understand that each target question contains at least one sub-question.

[0100] For each target question, identify the instruction verbs contained in it and obtain the corresponding text structure. Based on the instruction verbs and text structure, determine the question type for each target question.

[0101] Among them, the imperative verbs can include, but are not limited to, "solve", "prove", "choose", etc.

[0102] In one example, based on the instruction verbs and text structure, the target questions are categorized into question types such as fill-in-the-blank, multiple choice, true / false, solution, or proof.

[0103] Based on the above, assign corresponding question identifiers to each target question. Store the target questions based on the question identifiers.

[0104] Furthermore, for each target question, the question stem and reference answer are determined. Based on the established rules, the question stem and reference answer for each target question are standardized to obtain standardized target questions.

[0105] Corresponding to the above method embodiments, this application also provides embodiments of a question processing apparatus. Figure 3 A schematic diagram of the structure of a question processing device according to an embodiment of this application is shown. Figure 3 As shown, the device includes: The acquisition unit 302 is configured to acquire data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed.

[0106] The processing unit 304 is configured to input the question data to be processed and the prompt text into the question processing model to obtain at least one target question output by the question processing model. The question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question. Based on the question number and question type corresponding to each question, the hierarchical relationship between each question is determined to obtain at least one target question.

[0107] Furthermore, the processing unit 304 is further configured as follows: Determine the question storage format corresponding to the question data to be processed, and determine the target question data based on the question storage format; Based on the prompt text, the target question data is parsed to obtain at least one question, as well as the question number and question type corresponding to each question; Based on the question number and question type corresponding to each question, the hierarchical relationship between the questions is determined, and at least one target question is obtained.

[0108] Furthermore, the processing unit 304 is further configured as follows: Based on the question type, identify at least one question corresponding to each question type; Based on the question number, the hierarchical relationship between the questions is determined, wherein the hierarchical relationship includes the hierarchical relationship between questions of the same question type, and the hierarchical relationship between questions corresponding to different question types; Sort at least one question of the same question type based on the hierarchical relationship between the questions of the same question type to obtain at least one target question of the same question type. The questions corresponding to different question types are sorted according to the hierarchical relationship between the questions of different question types to obtain at least one target question corresponding to each question type.

[0109] Furthermore, the processing unit 304 is further configured as follows: Based on the prompt text, the target question data is parsed to obtain at least one question, the question number corresponding to each question, and the question semantics corresponding to each question are determined; Based on the semantics of each question, determine the instruction verbs corresponding to each question; Based on the instruction verbs corresponding to each question, determine the question type for each question.

[0110] Furthermore, the processing unit 304 is further configured as follows: If the question data to be processed is stored in image format, the image format is converted into text format to obtain the question data to be processed, and the target question data is obtained based on the question data to be processed and the set rules. If the question data to be processed is stored in text format, the text format is obtained based on the set rules to obtain the question data to be processed, and the target question data is obtained based on the question data to be processed and the set rules.

[0111] Furthermore, the processing unit 304 is further configured as follows: Obtain the semantic information corresponding to the question data to be processed; Based on the semantic information, the data context corresponding to the question data to be processed is determined; Based on the data context, the question data to be processed is modified to obtain the target question data.

[0112] Furthermore, the processing unit 304 is also configured as follows: Identify the target question to be processed, and obtain the target question stem and initial reference answer corresponding to the target question to be processed, wherein the target question to be processed is any one of the target questions; Determine the question stem attribute information corresponding to the target question stem; Based on the aforementioned question stem attribute information, a predicted reference answer is determined; Based on the predicted reference answer, the initial reference answer is adjusted to obtain the target reference answer.

[0113] Furthermore, the processing unit 304 is also configured as follows: Determine the initial target question, wherein the initial target question is any one of the target questions; Obtain the question representation format corresponding to the initial target question; Based on the preset format, the question representation format corresponding to the initial target question is adjusted to obtain the initial target question in the preset format.

[0114] The above is an illustrative scheme of a question processing device according to this embodiment. It should be noted that the technical solution of this question processing device and the technical solution of the question processing method described above belong to the same concept. For details not described in detail in the technical solution of the question processing device, please refer to the description of the technical solution of the question processing method described above.

[0115] Figure 4 A structural block diagram of a computing device according to an embodiment of this application is shown. The components of the computing device 400 include, but are not limited to, a memory 410 and a processor 420. The processor 420 is connected to the memory 410 via a bus 430, and a database 450 is used to store data.

[0116] The computing device 400 also includes an access device 440, which enables the computing device 400 to communicate via one or more networks 460. Examples of these networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 440 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Wi-MAX (Worldwide Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC) interface, and so on.

[0117] In one embodiment of this application, the aforementioned components of the computing device 400 and Figure 4 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 4 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this application. Those skilled in the art can add or replace other components as needed.

[0118] The computing device 400 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 400 can also be a mobile or stationary server.

[0119] The processor 420 is used to execute the following computer program / instruction, which, when executed by the processor, implements the steps of the above-mentioned problem-solving method.

[0120] The above is an illustrative scheme of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the above-described problem processing method belong to the same concept. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the above-described problem processing method.

[0121] An embodiment of this specification also provides a computer-readable storage medium storing a computer program / instructions that, when executed by a processor, implement the steps of the above-described problem-solving method.

[0122] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium and the technical solution of the above-described problem processing method belong to the same concept. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the above-described problem processing method.

[0123] An embodiment of this specification also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described problem-solving method.

[0124] The above is an illustrative scheme of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product and the technical solution of the above-described problem processing method belong to the same concept. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the above-described problem processing method.

[0125] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0126] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added or removed according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0127] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0128] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0129] The preferred embodiments disclosed above are merely illustrative of this application. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this application. These embodiments are selected and specifically described in this application to better explain the principles and practical applications of this application, thereby enabling those skilled in the art to better understand and utilize this application. This application is limited only by the claims and their full scope and equivalents.

Claims

1. A subject processing method characterized by, include: Obtain the data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed; The question data to be processed and the prompt text are input into the question processing model to obtain at least one target question output by the question processing model. The question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question. Based on the question number and question type corresponding to each question, the hierarchical relationship between each question is determined to obtain at least one target question.

2. The method of claim 1, wherein, Obtain at least one target question from the question processing model, including: Determine the question storage format corresponding to the question data to be processed, and determine the target question data based on the question storage format; Based on the prompt text, the target question data is parsed to obtain at least one question, as well as the question number and question type corresponding to each question; Based on the question number and question type corresponding to each question, the hierarchical relationship between the questions is determined, and at least one target question is obtained.

3. The method as described in claim 2, characterized in that, Based on the question number and question type corresponding to each question, the hierarchical relationship between the questions is determined, resulting in at least one target question, including: Based on the question type, identify at least one question corresponding to each question type; Based on the question number, the hierarchical relationship between the questions is determined, wherein the hierarchical relationship includes the hierarchical relationship between questions of the same question type, and the hierarchical relationship between questions corresponding to different question types; Sort at least one question of the same question type based on the hierarchical relationship between the questions of the same question type to obtain at least one target question of the same question type. The questions corresponding to different question types are sorted according to the hierarchical relationship between the questions of different question types to obtain at least one target question corresponding to each question type.

4. The method as described in claim 2, characterized in that, Based on the prompt text, the target question data is parsed to obtain at least one question, and the question number and question type corresponding to each question, including: Based on the prompt text, the target question data is parsed to obtain at least one question, the question number corresponding to each question, and the question semantics corresponding to each question are determined; Based on the semantics of each question, determine the instruction verbs corresponding to each question; Based on the instruction verbs corresponding to each question, determine the question type for each question.

5. The method as described in claim 2, characterized in that, Based on the aforementioned question storage format, the target question data is determined, including: If the question data to be processed is stored in image format, the image format is converted into text format to obtain the question data to be processed, and the target question data is obtained based on the question data to be processed and the set rules. If the question data to be processed is stored in text format, the text format is obtained based on the set rules to obtain the question data to be processed, and the target question data is obtained based on the question data to be processed and the set rules.

6. The method as described in claim 5, characterized in that, Based on the question data to be processed and the set rules, the target question data is obtained, including: Obtain the semantic information corresponding to the question data to be processed; Based on the semantic information, the data context corresponding to the question data to be processed is determined; Based on the data context, the question data to be processed is modified to obtain the target question data.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Identify the target question to be processed, and obtain the target question stem and initial reference answer corresponding to the target question to be processed, wherein the target question to be processed is any one of the target questions; Determine the question stem attribute information corresponding to the target question stem; Based on the aforementioned question stem attribute information, a predicted reference answer is determined; Based on the predicted reference answer, the initial reference answer is adjusted to obtain the target reference answer.

8. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Determine the initial target question, wherein the initial target question is any one of the target questions; Obtain the question representation format corresponding to the initial target question; Based on the preset format, the question representation format corresponding to the initial target question is adjusted to obtain the initial target question in the preset format.

9. A problem processing device, characterized in that, include: The acquisition unit is configured to acquire data of questions to be processed, wherein the data of questions to be processed includes at least one question to be processed; The processing unit is configured to input the question data to be processed and the prompt text into the question processing model to obtain at least one target question output by the question processing model. The question processing model is used to determine at least one question corresponding to the question to be processed based on the prompt text, as well as the question number and question type corresponding to each question. Based on the question number and question type corresponding to each question, the hierarchical relationship between each question is determined to obtain at least one target question.

10. A computing device, characterized in that, include: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions, which, when executed by the processor, implement the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium storing a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 8.

12. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 8.

Images