Information acquisition method and device, electronic equipment and computer readable storage medium

Abstract

Embodiments of the invention disclose an information acquisition method and apparatus, an electronic device and a computer readable storage medium. The method comprises the steps of obtaining original question and answer information, performing difficulty verification on the original question and answer information to obtain a difficulty verification result, and filtering the original question and answer information according to the difficulty verification result to obtain target question and answer information. By performing difficulty verification on the original question and answer information and filtering the original question and answer information based on the difficulty verification result, screening of the target question and answer information with relatively high quality is realized.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of information processing technology, and specifically to an information acquisition method, device, electronic device, and computer-readable storage medium. Background Art

[0002] With the rapid development of artificial intelligence (AI), optical character recognition (OCR) and visual question answering (VQA), as important research areas at the intersection of computer vision and natural language processing, have garnered widespread attention. Visual question answering systems, given an image and a question, automatically generate accurate responses by understanding the image content and question semantics. This technology holds broad application prospects in areas such as intelligent assistants, image retrieval, and assisting the visually impaired.

[0003] At present, visual question answering technology mainly trains models by constructing question and answer information (question and answer pair datasets), and question and answer information is mainly generated by extracting metadata corresponding to the original image and using question and answer templates based on the metadata.

[0004] However, since question and answer information affects the performance of the model, especially in complex scenarios, the model needs to be able to handle more complex problems. Therefore, higher requirements are placed on the question and answer information used to train the model. Therefore, how to determine the question and answer information used to train the model to improve the performance of the model in complex scenarios is an issue that needs to be solved urgently. Summary of the Invention

[0005] The embodiments of the present application provide an information acquisition method, device, electronic device, and computer-readable storage medium, which can filter out high-quality question and answer information.

[0006] In a first aspect, an embodiment of the present application provides an information acquisition method, the method comprising:

[0007] Get original question and answer information;

[0008] Performing difficulty verification on the original question and answer information to obtain a difficulty verification result;

[0009] The original question and answer information is filtered according to the difficulty verification result to obtain target question and answer information.

[0010] In a second aspect, an embodiment of the present application further provides an information acquisition device, the device comprising:

[0011] Acquisition module, used to obtain original question and answer information;

[0012] A verification module, configured to perform difficulty verification on the original question and answer information to obtain a difficulty verification result;

[0013] A filtering module is used to filter the original question and answer information according to the difficulty verification result to obtain target question and answer information.

[0014] Optionally, in some embodiments of the present application, performing difficulty verification on the original question and answer information to obtain a difficulty verification result includes:

[0015] Determine content validation model;

[0016] The original question and answer information is subjected to difficulty verification using the content verification model to obtain a difficulty verification result.

[0017] Optionally, in some embodiments of the present application, the difficulty verification result includes a preliminary verification result and a deep verification result;

[0018] The performing difficulty verification on the original question and answer information by the content verification model to obtain a difficulty verification result includes:

[0019] Performing preliminary difficulty verification on the original question and answer information using a content verification model within a first parameter range to obtain a preliminary verification result;

[0020] Preliminarily filtering the original question and answer information based on the preliminary verification result to obtain preliminary remaining question and answer information;

[0021] Performing a depth difficulty verification on the preliminary remaining question and answer information using a content verification model within a second parameter range to obtain a depth verification result;

[0022] The parameter value corresponding to the first parameter value range is smaller than the parameter threshold, and the interval span corresponding to the second parameter value range is larger than the interval threshold.

[0023] Optionally, in some embodiments of the present application, filtering the original question and answer information according to the difficulty verification result to obtain target question and answer information includes:

[0024] Filter the preliminary remaining question and answer information according to the deep verification result to obtain the target question and answer information,

[0025] or,

[0026] The original question and answer information is filtered according to the preliminary verification result and the deep verification result to obtain target question and answer information.

[0027] Optionally, in some embodiments of the present application, the original question-and-answer information records the image content of the original image;

[0028] The performing difficulty verification on the original question and answer information by the content verification model to obtain a difficulty verification result includes:

[0029] Extracting original question information and original answer information from the original question and answer information;

[0030] Outputting model answer information corresponding to the original question information based on the original image corresponding to the original question and answer information through the content verification model;

[0031] The difficulty verification result is determined according to the original response information and the model response information.

[0032] Optionally, in some embodiments of the present application, the original question-and-answer information records the image content of the original image;

[0033] After filtering the original question and answer information according to the difficulty verification result to obtain target question and answer information, the method further includes:

[0034] The original visual model is trained according to the target question and answer information and the original image corresponding to the target question and answer information to obtain a target visual model; wherein the target visual model is used to extract image content in the image to be identified.

[0035] Optionally, in some embodiments of the present application, before training the original visual model based on the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model, the method further includes:

[0036] Obtaining optimized question-and-answer information, where the optimized question-and-answer information is a manual review result obtained through a cross-validation strategy for the target question-and-answer information;

[0037] The training of the original visual model according to the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model includes:

[0038] The original visual model is trained according to the optimized question-answer information and the original image corresponding to the optimized question-answer information to obtain a target visual model.

[0039] In a third aspect, an embodiment of the present application further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps in the above-mentioned information acquisition method are implemented.

[0040] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above-mentioned information acquisition method are implemented.

[0041] In a fifth aspect, embodiments of the present application further provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various optional implementations described in the embodiments of the present application.

[0042] In an embodiment of the present application, original question and answer information is obtained, difficulty verification is performed on the original question and answer information to obtain a difficulty verification result, and the original question and answer information is filtered according to the difficulty verification result to obtain target question and answer information.

[0043] Among them, by verifying the difficulty of the original question and answer information and filtering the original question and answer information based on the difficulty verification result, the screening of high-quality target question and answer information is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in this application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 Schematic diagram of a scenario in which an electronic device according to an embodiment of the present application executes the information acquisition method;

[0046] Figure 2 Schematic diagram of the information acquisition method provided in the embodiment of the present application;

[0047] Figure 3 is a structural diagram of an information acquisition device provided in an embodiment of the present application;

[0048] Figure 4 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in this application to clearly and completely describe the technical solutions in this application. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0050] The embodiments of the present application provide an information acquisition method, device, electronic device and computer-readable storage medium. Specifically, the embodiments of the present application provide an information acquisition device suitable for electronic devices, for screening out high-quality question-and-answer information. Specifically, the electronic device includes a terminal device or a server, and the terminal device includes but is not limited to desktop computers, laptops, mobile phones, tablet computers and other devices. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks (CDN, Content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms. The server can be directly or indirectly connected via wired or wireless communication.

[0051] See also Figure 1 , Figure 1 : is a schematic diagram of a scenario in which an electronic device according to an embodiment of the present application performs the information acquisition method. The specific execution process of the electronic device performing the information acquisition method is as follows:

[0052] The electronic device 10 obtains original question and answer information, performs difficulty verification on the original question and answer information to obtain a difficulty verification result, and filters the original question and answer information according to the difficulty verification result to obtain target question and answer information.

[0053] For example, after the user receives the generated original question and answer information through an electronic device, he operates the electronic device or the electronic device automatically triggers the question and answer information screening process. Specifically: the electronic device performs difficulty verification on the original question and answer information to obtain a difficulty verification result, and filters the original question and answer information based on the difficulty verification result to obtain the target question and answer information.

[0054] In summary, the embodiment of the present application performs difficulty verification on the original question and answer information and filters the original question and answer information based on the difficulty verification result, thereby screening high-quality target question and answer information.

[0055] The following are detailed descriptions. It should be noted that the order of description of the following embodiments does not limit the priority order of the embodiments. Figure 2 , Figure 2 This is a flowchart of an information acquisition method provided in an embodiment of the present application. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order different from that shown in the flowchart. Specifically, the process of the information acquisition method specifically includes:

[0056] 101. Obtain original question and answer information.

[0057] In the embodiment of the present application, obtaining original question and answer information refers to obtaining information containing questions and corresponding answers, that is, question and answer pairs, from channels such as databases or the Internet. These original question and answer information can be questions asked by users on various platforms and their corresponding answers, or they can be information extracted from existing question and answer datasets. The original question and answer information usually contains a large amount of data. It is understandable that these original question and answer information contain both high-quality question and answer pairs and low-quality (e.g., unsuitable) question and answer pairs.

[0058] For example, raw question-and-answer information can be question-and-answer pairs about image content, including questions posed by users about the image and responses from other users or professionals, or questions about image content based on question templates and answers to those questions extracted from the image content using tools. These question-and-answer information may cover questions of varying difficulty levels, ranging from simple object recognition to complex scene understanding.

[0059] 102. Perform difficulty verification on the original question and answer information to obtain a difficulty verification result.

[0060] The difficulty verification result is an evaluation result of the difficulty of the original question and answer information, and the difficulty verification result includes high difficulty, medium difficulty and low difficulty.

[0061] In the embodiments of the present application, difficulty verification of the original question and answer information refers to evaluating the difficulty level of each question and answer information through a specific verification mechanism. The purpose of difficulty verification is to identify those question and answer information that meet the difficulty requirements and are valuable, while filtering out those question and answer information that are too simple or too complex.

[0062] Difficulty verification can be based on a variety of factors, such as the complexity of the question, the accuracy and completeness of the answer, the level of detail in the answer, the semantic relevance between the question and the answer, etc. Through a comprehensive evaluation of these factors, the difficulty verification result of each question and answer information can be obtained.

[0063] 103. Filter the original question and answer information according to the difficulty verification result to obtain the target question and answer information.

[0064] In the embodiment of the present application, filtering the original question and answer information according to the difficulty verification result means selecting the question and answer information that meets the specific difficulty requirements based on the aforementioned difficulty verification result to form the target question and answer information, thereby obtaining the OCR data used for visual model training.

[0065] In summary, the embodiment of the present application performs difficulty verification on the original question and answer information and filters the original question and answer information based on the difficulty verification result, thereby screening high-quality target question and answer information.

[0066] In the embodiments of the present application, the difficulty of the original question and answer information may be verified by a content verification model to obtain a difficulty verification result. That is, optionally, in some embodiments of the present application, the step of "performing difficulty verification on the original question and answer information to obtain a difficulty verification result" includes:

[0067] Determine content validation model;

[0068] The original question and answer information is verified for difficulty through the content verification model to obtain the difficulty verification result.

[0069] In an embodiment of the present application, the content verification model can be a model based on machine learning, such as a neural network model, a decision tree model, etc. The model evaluates the difficulty of the question and answer information by analyzing the semantic content, structural features, etc. of the question and answer information. Alternatively, the content verification model can also include a pre-trained language model, a knowledge graph model, or other models that can understand and evaluate the content of the question and answer. It should be noted that the choice of model depends on the characteristics of the question and answer information and the specific requirements of the verification, and the specific model selected is not limited here.

[0070] In the embodiments of the present application, the target question and answer information may be screened through a multi-round screening mechanism. That is, optionally, in some embodiments of the present application, the difficulty verification result includes a preliminary verification result and a deep verification result. The step of "performing difficulty verification on the original question and answer information through the content verification model to obtain a difficulty verification result" includes:

[0071] Performing preliminary difficulty verification on the original question and answer information through the content verification model within the first parameter range to obtain preliminary verification results;

[0072] The original question and answer information is preliminarily filtered through the preliminary verification results to obtain preliminary remaining question and answer information;

[0073] Performing a depth difficulty verification on the preliminary remaining question and answer information through a content verification model within the second parameter range to obtain a depth verification result;

[0074] The parameter value corresponding to the first parameter value range is smaller than the parameter threshold, and the interval span corresponding to the second parameter value range is larger than the interval threshold.

[0075] For example, a content verification model within a first parameter range is first used to perform a preliminary difficulty verification on the original question and answer information, obtaining a preliminary verification result. The parameter quantity corresponding to the first parameter range is less than a parameter threshold. For example, the parameter threshold can be set to 10 billion, keeping the parameter quantity of the content verification model within 10 billion. This lightweight model can quickly process large amounts of original question and answer information, improving verification efficiency. For example, models such as Qwen2-VL-7B, InternVL2.5-8B, and Idefics3-8B can be selected.

[0076] Then, the original question and answer information is preliminarily filtered based on the preliminary verification results to obtain preliminary residual question and answer information. Preliminary filtering can remove those question and answer information that obviously do not meet the requirements, such as overly simple questions or obviously wrong answers.

[0077] Next, the content verification model of the second parameter range is used to perform deep difficulty verification on the preliminary remaining question and answer information to obtain a deep verification result. Among them, the interval span corresponding to the second parameter range is greater than the interval threshold. For example, the parameter amount can be set to between 10 billion and 80 billion, and the interval threshold can be set to 70 billion, such as Qwen-VL-10B, Gemini1.0Nano, ChatGLM3-12B or PaLM-2S, etc. In addition, it can also be some closed-source models such as GPT-4o, Gemini-1.5-pro, etc. This large model has stronger understanding and judgment capabilities, and can perform more detailed difficulty assessments on question and answer information. At the same time, selecting models with larger parameter size differences for screening will help improve the objectivity and accuracy of the deep verification results.

[0078] Among them, preliminary difficulty verification and preliminary screening are first performed through models with smaller parameters, which effectively reduces the computational burden of large model reasoning during deep difficulty verification and improves the overall system resource utilization efficiency and operation speed.

[0079] Furthermore, when the difficulty verification result includes a preliminary verification result and a deep verification result, filtering can be performed in one of the following two ways, that is, the step of "filtering the original question and answer information according to the difficulty verification result to obtain the target question and answer information" includes:

[0080] Filter the preliminary remaining question and answer information according to the deep verification results to obtain the target question and answer information.

[0081] or,

[0082] The original question and answer information is filtered according to the preliminary verification results and the deep verification results to obtain the target question and answer information.

[0083] Among them, since the preliminary remaining question and answer information has been filtered based on the preliminary verification results, the preliminary remaining question and answer information can be directly filtered according to the deep verification results to obtain the target question and answer information with the required difficulty. Among them, in an embodiment of the present application, a difficulty threshold can be set, and the original question and answer information whose difficulty represented by the verification result is greater than the difficulty threshold is used as the target question and answer information. Among them, the difficulty threshold can be dynamically adjusted, for example, according to the feedback of the screening results.

[0084] In an embodiment of the present application, an answer to the original question information in the original question and answer information can be generated, and the generated answer can be compared with the original response information in the original question and answer information to determine the difficulty verification result corresponding to the original question and answer information.

[0085] Among them, in the embodiments of the present application, the answer corresponding to the original question information can be generated manually or by a model. For example, the model response information corresponding to the original question information in the original question and answer information is output by a content verification model, and the difficulty verification result of the original question and answer information is determined based on the comparison between the model response information and the original response information. For example, taking the original question and answer information as an example where the image content of the original image is recorded, the content verification model can generate the model response information for the original image based on the original question information in the original question and answer information, and determine the difficulty verification result of the original question and answer information based on the image content based on the comparison between the model response information and the original response information. That is, optionally, in some embodiments of the present application, the step of "verifying the difficulty of the original question and answer information through the content verification model to obtain the difficulty verification result" includes:

[0086] Extracting original question information and original answer information from original question and answer information;

[0087] Through the content verification model and the original image corresponding to the original question and answer information, the model answer information corresponding to the original question information is output;

[0088] The difficulty verification result is determined based on the original response information and the model response information.

[0089] When the original question-and-answer information records the image content of the original image, the original question information refers to the question posed regarding the image content, and the original answer information refers to the answer to the question based on the image content. The model answer information refers to the answer generated by the content verification model based on the image content and the original question information.

[0090] By comparing the differences between the original and model responses, we can assess the difficulty of the problem. For example, if the model response is highly consistent with the original, the problem is relatively easy for the model; if there are significant differences, the problem is relatively difficult for the model.

[0091] Accordingly, after obtaining target question-and-answer information that meets the difficulty requirement through screening, the original visual model can be trained based on the target temperature information to obtain a target visual model. For example, a target visual model that can more accurately extract text from an image can be obtained through training. Optionally, in some embodiments of the present application, the original question-and-answer information records the image content of the original image. After the step of "filtering the original question-and-answer information according to the difficulty verification result to obtain the target question-and-answer information", the method further includes:

[0092] The original visual model is trained according to the target question and answer information and the original image corresponding to the target question and answer information to obtain a target visual model; wherein the target visual model is used to extract image content in the image to be identified.

[0093] In addition, before model training, the target question and answer information can be optimized to obtain optimized question and answer information, and then the model can be trained based on the optimized question and answer information to improve the training effect of the model. That is, optionally, in some embodiments of the present application, before the step of "training the original visual model based on the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model", the method also includes:

[0094] Obtain optimized question and answer information. The optimized question and answer information is the manual review result obtained through the cross-validation strategy for the target question and answer information.

[0095] The original visual model is trained based on the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model, including:

[0096] The original visual model is trained based on the optimized question and answer information and the original images corresponding to the optimized question and answer information to obtain the target visual model.

[0097] Among them, the cross-validation strategy can include independent verification by multiple people, expert review, etc. to ensure the accuracy and quality of question and answer information. For example, for data that are still questionable after the first two rounds of automatic screening (referring to filtering based on preliminary filtering and deep verification results), experienced manual annotators are arranged to review it. Each piece of data must be checked one by one by two annotators based on the image and metadata, focusing on comparing the correspondence between the text displayed in the image and the question and answer. For data with inconsistent annotation results from the two annotators, a third annotator is arranged to perform cross-validation to ensure the accuracy and consistency of the data content. Annotators flexibly allocate annotation tasks based on the total amount of data. For example, each annotator is responsible for approximately 300 pieces of data.

[0098] Among them, the performance and generalization ability of the model can be further improved by using manually optimized question and answer information for training.

[0099] Understandably, since the selected target Q&A information or optimized Q&A information is of a higher difficulty type, training the model based on this target Q&A information helps improve the trained model's ability to solve complex problems and enhance its performance when faced with complex problems. The optimized Q&A information obtained through manual review ensures consistency with the content in the original image, helping to improve model training results and enhance the trained model's image recognition accuracy.

[0100] Among them, in the embodiment of the present application, the preliminary difficulty verification, deep difficulty verification and manual review can be carried out in sequence, and the screening corresponding to the preliminary difficulty verification and deep difficulty verification can greatly reduce the workload of manual verification, and only a small amount of difficult data needs to be manually reviewed. Compared with the traditional solution that requires each original question and answer information to be checked one by one, the embodiment of the present application reduces labor costs and time investment.

[0101] To facilitate better implementation of the information acquisition method of the present application, the present application also provides an information acquisition device based on the above information acquisition method. The meanings of the terms are the same as those in the above information acquisition method, and the specific implementation details can be referred to the description in the method embodiment.

[0102] See also Figure 3 , Figure 3 : is a schematic diagram of the structure of the information acquisition device provided in an embodiment of the present application. The information acquisition device may be specifically as follows:

[0103] Acquisition module 201, used to obtain original question and answer information;

[0104] Verification module 202, used to perform difficulty verification on the original question and answer information to obtain a difficulty verification result;

[0105] The filtering module 203 is used to filter the original question and answer information according to the difficulty verification result to obtain the target question and answer information.

[0106] Optionally, in some embodiments of the present application, difficulty verification is performed on the original question and answer information to obtain a difficulty verification result, including:

[0107] Determine content validation model;

[0108] The original question and answer information is verified for difficulty through the content verification model to obtain the difficulty verification result.

[0109] Optionally, in some embodiments of the present application, the difficulty verification result includes a preliminary verification result and a depth verification result;

[0110] The original question and answer information is verified for difficulty through the content verification model to obtain difficulty verification results, including:

[0111] Performing preliminary difficulty verification on the original question and answer information through the content verification model within the first parameter range to obtain preliminary verification results;

[0112] The original question and answer information is preliminarily filtered through the preliminary verification results to obtain preliminary remaining question and answer information;

[0113] Performing a depth difficulty verification on the preliminary remaining question and answer information through a content verification model within the second parameter range to obtain a depth verification result;

[0114] The parameter value corresponding to the first parameter value range is smaller than the parameter threshold, and the interval span corresponding to the second parameter value range is larger than the interval threshold.

[0115] Optionally, in some embodiments of the present application, the original question and answer information is filtered according to the difficulty verification result to obtain the target question and answer information, including:

[0116] Filter the preliminary remaining question and answer information according to the deep verification results to obtain the target question and answer information.

[0117] or,

[0118] The original question and answer information is filtered according to the preliminary verification results and the deep verification results to obtain the target question and answer information.

[0119] Optionally, in some embodiments of the present application, the original question-and-answer information records the image content of the original image;

[0120] The original question and answer information is verified for difficulty through the content verification model to obtain difficulty verification results, including:

[0121] Extracting original question information and original answer information from original question and answer information;

[0122] Through the content verification model and the original image corresponding to the original question and answer information, the model answer information corresponding to the original question information is output;

[0123] The difficulty verification result is determined based on the original response information and the model response information.

[0124] Optionally, in some embodiments of the present application, the original question-and-answer information records the image content of the original image;

[0125] After filtering the original question and answer information according to the difficulty verification result to obtain the target question and answer information, the method further includes:

[0126] The original visual model is trained according to the target question and answer information and the original image corresponding to the target question and answer information to obtain a target visual model; wherein the target visual model is used to extract image content in the image to be identified.

[0127] Optionally, in some embodiments of the present application, before training the original visual model based on the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model, the method further includes:

[0128] Obtain optimized question and answer information. The optimized question and answer information is the manual review result obtained through the cross-validation strategy for the target question and answer information.

[0129] The original visual model is trained based on the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model, including:

[0130] The original visual model is trained based on the optimized question and answer information and the original images corresponding to the optimized question and answer information to obtain the target visual model.

[0131] In the embodiment of the present application, the acquisition module 201 acquires the original question and answer information, the verification module 202 performs difficulty verification on the original question and answer information to obtain a difficulty verification result, and the filtering module 203 filters the original question and answer information according to the difficulty verification result to obtain target question and answer information.

[0132] Among them, the embodiment of the present application verifies the difficulty of the original question and answer information and filters the original question and answer information based on the difficulty verification result to achieve the screening of target question and answer information with higher quality.

[0133] In addition, the present application also provides an electronic device, such as Figure 4 As shown, it shows a schematic diagram of the structure of the electronic device involved in this application, specifically:

[0134] The electronic device may include one or more processors 301 of processing cores, one or more computer-readable storage media memories 302, a power supply 303, an input unit 304 and other components. Those skilled in the art will appreciate that Figure 4 The electronic device structure shown in the figure does not constitute a limitation of the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently.

[0135] The processor 301 is the control center of the electronic device. It connects all parts of the electronic device using various interfaces and lines. By running or executing software programs and / or modules stored in the memory 302 and accessing data stored in the memory 302, it performs various functions of the electronic device and processes data, thereby monitoring the electronic device as a whole. Optionally, the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor primarily processes the operating system, user interface, and application programs, while the modem processor primarily handles wireless communications. It is understood that the modem processor may not be integrated into the processor 301.

[0136] The memory 302 can be used to store software programs and modules. The processor 301 executes various functional applications and data processing by running the software programs and modules stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 302 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory 302 may also include a memory controller to provide the processor 301 with access to the memory 302.

[0137] The electronic device also includes a power supply 303 for supplying power to various components. Preferably, the power supply 303 can be logically connected to the processor 301 via a power management system, thereby enabling the power management system to manage charging, discharging, and power consumption. The power supply 303 can also include one or more DC or AC power supplies, a recharging system, a power supply device debugging circuit, a power converter or inverter, a power status indicator, and other arbitrary components.

[0138] The electronic device may further include an input unit 304, which may be configured to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0139] Although not shown, the electronic device may further include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 301 in the electronic device will load the executable files corresponding to the processes of one or more application programs into the memory 302 according to the following instructions, and the processor 301 will run the application programs stored in the memory 302, thereby implementing the steps of any of the information acquisition methods provided in the embodiments of the present application.

[0140] In an embodiment of the present application, original question and answer information is obtained, difficulty verification is performed on the original question and answer information to obtain a difficulty verification result, and the original question and answer information is filtered according to the difficulty verification result to obtain target question and answer information.

[0141] Among them, by verifying the difficulty of the original question and answer information and filtering the original question and answer information based on the difficulty verification result, the screening of high-quality target question and answer information is achieved.

[0142] The specific implementation of the above operations can be found in the previous embodiments and will not be repeated here.

[0143] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be accomplished by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0144] To this end, the present application provides a computer-readable storage medium, on which a computer program is stored. The computer program can be loaded by a processor to execute the steps in any information acquisition method provided in the present application.

[0145] The specific implementation of the above operations can be found in the previous embodiments and will not be repeated here.

[0146] The computer-readable storage medium may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0147] Since the instructions stored in the computer-readable storage medium can execute the steps in any information acquisition method provided in the present application, the beneficial effects that can be achieved by any information acquisition method provided in the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0148] The above is a detailed introduction to the information acquisition method, device, electronic device and computer-readable storage medium provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. An information acquisition method, characterized in that: The method comprises: Get original question and answer information; Performing difficulty verification on the original question and answer information to obtain a difficulty verification result; The original question and answer information is filtered according to the difficulty verification result to obtain target question and answer information.

2. The information acquisition method according to claim 1, characterized in that The difficulty verification of the original question and answer information to obtain a difficulty verification result includes: Determine content validation model; The original question and answer information is subjected to difficulty verification using the content verification model to obtain a difficulty verification result.

3. The information acquisition method according to claim 2, characterized in that: The difficulty verification results include preliminary verification results and in-depth verification results; The performing difficulty verification on the original question and answer information by the content verification model to obtain a difficulty verification result includes: Performing preliminary difficulty verification on the original question and answer information using a content verification model within a first parameter range to obtain a preliminary verification result; Preliminarily filtering the original question and answer information based on the preliminary verification result to obtain preliminary remaining question and answer information; Performing a depth difficulty verification on the preliminary remaining question and answer information using a content verification model within a second parameter range to obtain a depth verification result; The parameter value corresponding to the first parameter value range is smaller than the parameter threshold, and the interval span corresponding to the second parameter value range is larger than the interval threshold.

4. The information acquisition method according to claim 3, characterized in that: The filtering of the original question and answer information according to the difficulty verification result to obtain target question and answer information includes: Filter the preliminary remaining question and answer information according to the deep verification result to obtain the target question and answer information, or, The original question and answer information is filtered according to the preliminary verification result and the deep verification result to obtain target question and answer information.

5. The information acquisition method according to claim 2, characterized in that: The original question-and-answer information records the image content of the original image; The performing difficulty verification on the original question and answer information by the content verification model to obtain a difficulty verification result includes: Extracting original question information and original answer information from the original question and answer information; Outputting model answer information corresponding to the original question information based on the original image corresponding to the original question and answer information through the content verification model; The difficulty verification result is determined according to the original response information and the model response information.

6. The information acquisition method according to claim 1, wherein: The original question-and-answer information records the image content of the original image; After filtering the original question and answer information according to the difficulty verification result to obtain target question and answer information, the method further includes: The original visual model is trained according to the target question and answer information and the original image corresponding to the target question and answer information to obtain a target visual model; wherein the target visual model is used to extract image content in the image to be identified.

7. The information acquisition method according to claim 6, characterized in that: Before training the original visual model according to the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model, the method further includes: Obtaining optimized question-and-answer information, where the optimized question-and-answer information is a manual review result obtained through a cross-validation strategy for the target question-and-answer information; The training of the original visual model according to the target question and answer information and the original image corresponding to the target question and answer information to obtain the target visual model includes: The original visual model is trained according to the optimized question-answer information and the original image corresponding to the optimized question-answer information to obtain a target visual model.

8. An information acquisition device, characterized in that: The device comprises: Acquisition module, used to obtain original question and answer information; A verification module, configured to perform difficulty verification on the original question and answer information to obtain a difficulty verification result; A filtering module is used to filter the original question and answer information according to the difficulty verification result to obtain target question and answer information.

9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the information acquisition method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the information acquisition method according to any one of claims 1 to 7 are implemented.

Images