Library updating, question answering and resource sharing method and device based on large model, and information display method and device

Abstract

The invention provides a library updating, question answering and resource sharing method and device based on a large model and an information display method and device, relates to the technical field of artificial intelligence, in particular to the technical fields of deep learning, large models and the like, and can be used for application scenes such as generative search, intelligent document editing, intelligent assistants and virtual assistants. The library updating method based on the large model comprises the steps that in response to a received updating request, for multiple files of a target object included in the updating request, the object type of the target object is determined according to candidate summary information of the multiple files, and the multiple files have multiple file types; according to a preset template of the object type, processing the multiple pieces of candidate summary information to obtain object information; and updating the library according to the object information, the plurality of files and the respective candidate summary information of the plurality of files.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, in particular to technical fields such as deep learning and big models, and can be used in application scenarios such as generative search, intelligent document editing, intelligent assistants, virtual assistants, and more specifically, to a library update, question and answer, resource sharing method, information display method and device based on big models. Background Art

[0002] With the development of computer technology and network technology, the Internet is being used more and more widely in various fields, and the amount of global information is also increasing dramatically, generating different types of multimodal data such as text, images, audio and video. Summary of the invention

[0003] The present invention provides a library update, question and answer, resource sharing method, information display method and device based on a large model.

[0004] According to one aspect of the present disclosure, a large model-based library update method is provided, comprising: in response to receiving an update request, for multiple files of a target object included in the update request, determining an object type of the target object according to candidate summary information of each of the multiple files, wherein the multiple files have multiple file types; processing the multiple candidate summary information according to a preset template for the object type to obtain object information; and updating the library according to the object information, the multiple files and the candidate summary information of each of the multiple files.

[0005] According to one aspect of the present disclosure, a question-answering method based on a big model is provided, comprising: in response to receiving a question-answering request, determining a target library to be matched according to the intention of the question text in the question-answering request; matching the target library according to the question text to obtain a matching result, wherein the target library is obtained by using a library updating method based on a big model; and determining the answer text according to the matching result using the big model.

[0006] According to one aspect of the present disclosure, a resource sharing method based on a big model is provided, comprising: in response to receiving a sharing request, processing summary information of the target object selected in a library indicated by the sharing request to obtain processed information, wherein the library is obtained by using a library update method based on a big model; and resource sharing of the processed information.

[0007] According to one aspect of the present disclosure, there is provided an information display method, comprising: in response to receiving a user request, sending a pending request, wherein the pending request includes at least one of an update request, a question and answer request, and a sharing request; and in response to obtaining response information generated in response to the pending request, displaying the response information, wherein the response information is generated by using a library update method based on a big model, a question and answer method based on a big model, or a resource sharing method based on a big model, and the response information includes at least one of object information, answer text, and processed information.

[0008] According to another aspect of the present disclosure, a library updating device based on a large model is provided, comprising: a first determining module, for determining, in response to receiving an update request, an object type of the target object according to candidate summary information of each of the multiple files of the target object included in the update request, wherein the multiple files have multiple file types; a first processing module, for processing the multiple candidate summary information according to a preset template for the object type to obtain object information; and an updating module, for updating the library according to the object information, the multiple files and the candidate summary information of each of the multiple files.

[0009] According to another aspect of the present disclosure, a question-and-answer device based on a big model is provided, comprising: a second determination module, for determining, in response to receiving a question-and-answer request, a target library to be matched according to the intention of the question text in the question-and-answer request; a matching module, for matching the target library according to the question text to obtain a matching result, wherein the target library is obtained by using a library updating device based on a big model; and a third determination module, for determining the answer text according to the matching result using the big model.

[0010] According to another aspect of the present disclosure, a resource sharing device based on a big model is provided, comprising: a second processing module, for, in response to receiving a sharing request, processing summary information of the target object selected in a library indicated by the sharing request to obtain processed information, wherein the library is obtained by using a library updating device based on a big model; and a resource sharing module, for performing resource sharing on the processed information.

[0011] According to another aspect of the present disclosure, there is provided an information display device, comprising: a sending module, for sending a pending request in response to receiving a user request, wherein the pending request comprises at least one of an update request, a question and answer request and a sharing request; and a display module, for displaying the response information in response to obtaining response information generated in response to the pending request, wherein the response information is generated by using a library updating device based on a large model, a question and answer device based on a large model or a resource sharing device based on a large model, and the response information comprises at least one of object information, answer text and processed information.

[0012] According to another aspect of the present disclosure, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0013] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, the steps of the above method are implemented.

[0014] According to another aspect of the present disclosure, a computer program product is provided, including a computer program or instructions, and the steps of the above method are implemented when the computer program or instructions are executed by a processor.

[0015] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0017] Figure 1 The system architecture of the library update, question and answer, resource sharing method and information display method based on the big model according to the embodiment of the present disclosure is schematically shown;

[0018] Figure 2 A flowchart of a library updating method based on a large model according to an embodiment of the present disclosure is schematically shown;

[0019] Figure 3 An example schematic diagram schematically shows a process of uploading multiple files according to an embodiment of the present disclosure;

[0020] Figure 4 An example schematic diagram schematically shows a process of determining an object name and a file name according to an embodiment of the present disclosure;

[0021] Figure 5A An example schematic diagram schematically shows directory information according to an embodiment of the present disclosure;

[0022] Figure 5B An example schematic diagram schematically shows summary information according to an embodiment of the present disclosure;

[0023] Figure 6 An example schematic diagram of a library update process based on a large model according to an embodiment of the present disclosure is schematically shown;

[0024] Figure 7 The flowchart of the question-answering method based on the big model according to the embodiment of the present disclosure is schematically shown;

[0025] Figure 8 An example schematic diagram of a question-answering process based on a large model according to an embodiment of the present disclosure is schematically shown;

[0026] Fig. 9 A flowchart of a resource sharing method based on a large model according to an embodiment of the present disclosure is schematically shown;

[0027] Fig.10 An example schematic diagram schematically shows a resource sharing process according to an embodiment of the present disclosure;

[0028] Fig.11 A flowchart of an information display method according to an embodiment of the present disclosure is schematically shown;

[0029] Fig.12 A block diagram of a library updating device based on a large model according to an embodiment of the present disclosure is schematically shown;

[0030] Fig.13 A block diagram of a question-answering device based on a large model according to an embodiment of the present disclosure is schematically shown;

[0031] Fig.14 A block diagram of a resource sharing device based on a large model according to an embodiment of the present disclosure is schematically shown;

[0032] Fig.15 A block diagram of an information display device according to an embodiment of the present disclosure is schematically shown;

[0033] Fig.16 A block diagram schematically shows a structure of an intelligent agent of a large model according to an embodiment of the present disclosure; and

[0034] Fig.17 A block diagram of an electronic device suitable for implementing library update, question and answer, resource sharing method, and information display method based on a large model according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0035] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0036] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0037] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0038] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0039] With the development of computer technology and network technology, the Internet is being used more and more widely in various fields, and the amount of global information is also increasing dramatically. For example, in the legal field, different types of cases have large-scale multimodal data of different file types.

[0040] In one example, AI products in the legal field can include products that provide legal services to ordinary users and products that provide efficiency tools to law firms / lawyers, depending on the user groups they target. AI products for law firms / lawyers usually focus on the case handling process in the case workflow, but lack products related to the case discussion process. Therefore, how to obtain differentiated AI products from the case discussion process to assist users in sorting out documents and finding the required information quickly and effectively is an urgent problem to be solved.

[0041] To this end, the embodiment of the present disclosure proposes a library update solution based on a large model. For example, in response to receiving an update request, for multiple files of a target object included in the update request, the object type of the target object is determined according to candidate summary information of each of the multiple files, wherein the multiple files have multiple file types; the multiple candidate summary information is processed according to a preset template for the object type to obtain object information; and the library is updated according to the object information, the multiple files, and the candidate summary information of each of the multiple files.

[0042] According to an embodiment of the present disclosure, by using candidate summary information of each of multiple files to determine the object type of the target object, it is possible to achieve intelligent recognition of multimodal files, and improve the processing efficiency of files of different file types. By processing multiple candidate summary information according to a preset template for the object type, the object information can be automatically obtained, and differentiated processing for different object types can be achieved, which can ensure the accuracy of the object information. On this basis, by updating the library according to the object information, multiple files, and the candidate summary information of each of the multiple files, a private domain library can be intelligently constructed, and the efficiency and accuracy of library updates can be improved.

[0043] In the technical solution of the present invention, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0044] In the technical solution of the present invention, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.

[0045] Figure 1 The schematic diagram shows the system architecture that can apply the library update, question and answer, resource sharing method, and information display method based on the big model according to the embodiment of the present disclosure. It should be noted that Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0046] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104 and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0047] The user may use at least one of the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0048] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0049] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0050] It should be noted that the library update, question and answer, resource sharing method, and information display method based on the big model provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the library update, question and answer, resource sharing device, and information display device based on the big model provided in the embodiment of the present disclosure can generally be set in the server 105. The library update, question and answer, resource sharing method, and information display method based on the big model provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or the server 105. Accordingly, the library update, question and answer, resource sharing device, and information display device based on the big model provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or the server 105.

[0051] Alternatively, the library update, question and answer, resource sharing method, and information display method based on the large model provided in the embodiments of the present disclosure may also be executed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may also be executed by other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103. Accordingly, the library update, question and answer, resource sharing device, and information display device based on the large model provided in the embodiments of the present disclosure may also be set in the first terminal device 101, the second terminal device 102, or the third terminal device 103, or in other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0052] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0053] It should be noted that the sequence numbers of the operations in the following method are only used as representations of the operations for the purpose of description, and should not be regarded as representing the execution order of the operations. Unless explicitly stated, the method does not need to be executed completely in the order shown.

[0054] The above describes the system architecture of the library update, question and answer, resource sharing method, and information display method provided by the present disclosure, which can be applied based on the large model. Figure 2 For example, the large model-based library update process of the present disclosure is further described.

[0055] Figure 2 The flowchart of the library updating method based on the large model according to the embodiment of the present disclosure is schematically shown.

[0056] like Figure 2 As shown, the large model-based library updating method 200 includes operations S210 to S230.

[0057] In operation S210 , in response to receiving an update request, for a plurality of files of a target object included in the update request, an object type of the target object is determined according to candidate summary information of each of the plurality of files.

[0058] In operation S220, a plurality of candidate summary information is processed according to a preset template of the object type to obtain object information.

[0059] In operation S230, a library is updated based on the object information, the plurality of files, and the candidate summary information of each of the plurality of files.

[0060] The multiple files may be multi-modal files for the same target object. A multi-modal file refers to a file containing multiple data forms, that is, the multiple files may have multiple file types. For example, the file type may include at least one of the following: text, audio, picture, video, and chart, etc.

[0061] For each file, the file can be processed according to the file type and a method corresponding to the file type to obtain candidate summary information. The candidate summary information can refer to information obtained by briefly refining key information in the file. The processing method for different file types can be configured according to actual business needs and is not limited here.

[0062] For example, for text files, you can directly extract the content and perform keyword analysis or semantic understanding to obtain candidate summary information. Alternatively, for audio files, you can use speech recognition technology to convert the audio into text, and then generate candidate summary information based on the text information. Alternatively, for image files, you can extract visual features through image recognition algorithms, and generate candidate summary information based on the visual features. Alternatively, for video files, you can combine image and audio analysis to extract key frames, keywords, and speech content to generate candidate summary information.

[0063] After obtaining candidate summary information of each of the multiple files, the object type of the target object can be determined according to the candidate summary information of each of the multiple files. For example, in the legal field, the target object can refer to a legal case, and the object type of the target object can refer to a case type.

[0064] Multiple candidate object types and candidate preset templates for each of the multiple candidate object types may be preconfigured. The candidate preset template may refer to a document design strategy for automatically generating a corresponding template structure for data of a specific candidate object type. After determining the object type of the target object, a preset template corresponding to the object type may be determined from the multiple candidate preset templates based on the object type and the multiple candidate object types.

[0065] For example, the preset template may include multiple preset fields. For each preset field in the preset template for the object type, a field value corresponding to the preset field may be determined in the candidate summary information based on the preset field. On this basis, the extracted field value may be filled into the corresponding preset field position in the preset template to obtain structured object information.

[0066] The specific form of the preset template can be configured according to actual business needs and is not limited here. For example, the preset template may include a template for generating a directory and a template for generating a summary. The template for generating a directory may include information for organizing and outlining the core content in the target object, which can help quickly browse and find the various parts of the target object. The template for generating a summary may include information summarizing the overall overview and main conclusions of the target object, which can help quickly understand the substantive content and main conclusions of the target object.

[0067] In one example, a template for generating a directory corresponding to the object type can be determined from multiple candidate preset templates according to the object type. On this basis, multiple candidate summary information can be filled into the template for generating a directory to obtain directory information. For example, a field value corresponding to a preset field can be extracted from the candidate summary information, and the extracted field value can be filled into the corresponding preset field position in the template for generating a directory to obtain target information.

[0068] In another example, a template for generating a summary corresponding to the object type may be determined from a plurality of candidate preset templates according to the object type. On this basis, the plurality of candidate summary information may be filled into the template for generating a summary to obtain summary information. For example, a field value corresponding to a preset field may be extracted from the candidate summary information, and the extracted field value may be filled into a corresponding preset field position in the template for generating a summary to obtain summary information.

[0069] After obtaining the object information, the library can be updated according to the object information, the plurality of files and the candidate summary information of each of the plurality of files. For example, the library can include a first database and a second database, the first database can be used to store information related to the target object, and the second database can be used to store information related to the files for the target object.

[0070] According to an embodiment of the present disclosure, by using candidate summary information of each of multiple files to determine the object type of the target object, it is possible to achieve intelligent recognition of multimodal files, and improve the processing efficiency of files of different file types. By processing multiple candidate summary information according to a preset template for the object type, the object information can be automatically obtained, and differentiated processing for different object types can be achieved, which can ensure the accuracy of the object information. On this basis, by updating the library according to the object information, multiple files, and the candidate summary information of each of the multiple files, a private domain library can be intelligently constructed, and the efficiency and accuracy of library updates can be improved.

[0071] The above describes the library update process based on the large model. Figure 3 The process of uploading multiple files is schematically described.

[0072] Figure 3 An example schematic diagram of a multiple file upload process according to an embodiment of the present disclosure is schematically shown.

[0073] like Figure 3 As shown in 300 , the multiple files of the target object may include, for example, an audio type file 301 , a picture type file 302 , a text type file 303 , a video type or a mixed type file 304 , and the like.

[0074] Click or drag multiple files of different file types for the target object to the front-end interface 305, and after completing the click or drag of multiple files, click the start upload control 306 of the front-end interface to achieve batch uploading of multiple files for the target object.

[0075] After receiving the multiple files included in the update request, for each of the multiple files, a summary generation tool can be determined based on the file type of the file. The summary generation tool is used to process files of different file types to obtain candidate summary information. On this basis, the summary generation tool can be called using the large model. Based on the summary generation tool, the file is processed to obtain candidate summary information of the file.

[0076] For example, for a text-type file 303, the relationship between the length of the file 303 and the threshold value can be first determined. When the length of the file 303 is less than or equal to the threshold value, the file 303 can be directly determined as candidate summary information. When the file 303 is greater than the threshold value, the summary generation tool corresponding to the text type can be used to process the file 303 to obtain candidate summary information. The threshold value can be configured according to actual business needs and is not limited here. For example, it can be positively correlated with the input processing length of the large model.

[0077] Alternatively, for the audio type file 301, the first preset model can be used to process the file 301 to obtain the text of the file 301. The first preset model can be configured according to actual business needs and is not limited here. For example, the first preset model can be a speech recognition model or a speaker recognition model. For the text of the file 301, the operation corresponding to the text type file is performed.

[0078] Alternatively, for the image type file 302, the second preset model can be used to process the file 302 to obtain the text of the file 302. The second preset model can be configured according to actual business needs and is not limited here. For example, the second preset model can be an image understanding model and an optical character recognition model. For the text of the file 302, the operation corresponding to the text type file is performed.

[0079] Alternatively, for a video type or mixed type file 304, the file 304 may be split into a text type file, an audio type file, and an image type file, and the above-mentioned processing methods for files of different file types are used to process each type of file separately to obtain sub-summary information of each part. On this basis, candidate summary information may be generated according to the order relationship between the files obtained by splitting the file 304 and the sub-summary information of each file.

[0080] According to the embodiments of the present disclosure, through the combination of a large model and a summary generation tool, the automation, flexibility and intelligence of multimodal document summary generation is achieved, the efficiency of large-scale data processing and user experience are improved, and the accuracy and scalability of multimodal data processing are enhanced.

[0081] The above describes the process of uploading multiple files. Figure 4 A schematic description of the process of determining object names and file names is given.

[0082] Figure 4 An example schematic diagram of a process of determining an object name and a file name according to an embodiment of the present disclosure is schematically shown.

[0083] like Figure 4 As shown, in 400, the front-end interface 401 can scan and access the back-end in real time, query the generation status, and display a progress bar. For example, the progress bar can include a file upload step, an intelligent analysis step, and an intelligent summary step. After the user uploads the file, the file upload step in the progress bar can be presented as completed.

[0084] In one example, the candidate summary information can be parsed to obtain keywords in the candidate summary information, and the intent can be determined based on the keywords in the candidate summary information. The intent is to help quickly understand the main content, key information and core theme of the file by summarizing the key information of the file or content.

[0085] After obtaining the intent of the candidate summary information, the plurality of candidate summary information may be classified according to the intent of each candidate summary information, so as to determine a plurality of target summary information belonging to a target type among the plurality of candidate summary information, thereby filtering a specific type of target summary information among the plurality of candidate summary information. The target type may refer to a specific type of file, for example, in the legal field, the target type may refer to an interview type of file.

[0086] On this basis, the large model can be used to classify the target object according to multiple target summary information belonging to the target type to obtain the object type. For example, a preset method can be used to process multiple target summary information to accurately determine the object type. The preset method can be configured according to actual business needs and is not limited here. For example, the preset method can include at least one of the following: rule-based matching, semantic analysis-based matching, Naive Bayes (NB), Support Vector Machine (SVM) or deep learning model, etc.

[0087] According to the embodiments of the present disclosure, the analysis efficiency and intelligence level of multimodal data are improved by performing intent recognition on candidate summary information and combining it with the automated processing of a large model. On this basis, by screening out target summary information belonging to the target type from multiple candidate summary information, and using the large model to determine the object type based on the target summary information, the data processing and classification process is simplified, and the accuracy of object type determination is improved.

[0088] After obtaining the object type, for the target object, an object name can also be generated based on the object type and multiple target summary information. In one example, a large model can be used to extract elements such as participants, time, and location from multiple target summary information, and the object name can be intelligently named in combination with the object type. For example, if the participant is Ms. Zhang and the object type is type A, the object name can be generated as "Ms. Zhang_Type A".

[0089] In addition, for each file, a file name can be generated based on the object type, file type, and candidate summary information. In one example, a large model can be used to extract elements such as participants, time, and location from the candidate summary information, and the file name can be intelligently named in combination with the object type. For example, if the participant is Ms. Zhang, the object type is type A, and the file type is "picture type", the file name can be generated as "Ms. Zhang_Type A_Picture".

[0090] like Figure 4 As shown, the front-end interface 401 can display the names of the intelligently named files, for example, "X_X_Interview Recording", "X_X_Picture", and "X_X_Chat Record", etc. In addition, the rename control 402 provided by the front-end interface 401 can be used to rename the files. After the object name and the file name are generated, the intelligent analysis step in the progress bar can be presented as completed.

[0091] According to the embodiments of the present disclosure, by using a large model, an object name is intelligently generated according to an object type and multiple target summary information, thereby improving the intelligence and accuracy of object name generation. In addition, by using a large model, a file name is intelligently generated according to an object type and candidate summary information, thereby improving the intelligence and accuracy of file name generation.

[0092] The above describes the process of determining the object name and the file name. The following will use 5A to schematically describe the directory information.

[0093] Figure 5A An example schematic diagram of directory information according to an embodiment of the present disclosure is schematically shown.

[0094] like Figure 5A As shown, the preset template may include a template for generating a directory, and correspondingly, the object information may include directory information. Templates for generating a directory corresponding to different object types may be preconfigured. When generating directory information 510, the large model selects a corresponding template for generating a directory according to the object type. The large model generates a directory to be filled based on the template for generating a directory, and classifies and organizes each file according to the file names of each of the multiple files, and intelligently generates directory information 510.

[0095] In one example, the directory information 510 may include different folder categories corresponding to the object name A, each folder category having files of a corresponding category. For example, folder A includes file "X_X_A", folder B includes file "X_X_B" and file "X_X_C", folder C includes file "X_X_D", and so on.

[0096] The above is a schematic description of the directory information. Figure 5B Provide a schematic description of the summary information.

[0097] Figure 5B An example schematic diagram of summary information according to an embodiment of the present disclosure is schematically shown.

[0098] like Figure 5B As shown, the preset template may include a template for generating a summary, and correspondingly, the object information may include summary information. Templates for generating summaries corresponding to different object types may be preconfigured. When generating summary information 520, the large model selects a corresponding template for generating a summary according to the object type. The large model generates a summary to be filled based on the template for generating a summary, and performs a structured summary of the target object according to multiple target summary information and directory information, and intelligently generates summary information 520.

[0099] In one example, the summary information 520 may include basic information, object introduction, main demands, and a list of documents.

[0100] According to the embodiments of the present disclosure, by using a large model combined with a preset template for an object type, the automatic generation of directory information and summary information is achieved, and the efficiency and accuracy of multimodal data processing are improved. Since the content of different object types can be uniformly and standardized through their respective preset templates, the adaptability of the generation of directory information and summary information is improved.

[0101] The above is a schematic description of the directory information and summary information. Figure 6 The library update process based on the large model is schematically described.

[0102] Figure 6 An example schematic diagram of a library update process based on a large model according to an embodiment of the present disclosure is schematically shown.

[0103] like Figure 6 As shown, in 600, for each file 601 of the plurality of files of the target object, candidate summary information 602 of the file 601 may be determined. After the candidate summary information 602 of each of the plurality of files 601 is obtained, operation S610 may be performed.

[0104] In operation S610, it can be determined whether the candidate summary information 602 belongs to the target type. For example, multiple target summary information 603 belonging to the target type can be determined from multiple candidate summary information 602 according to the intention of each candidate summary information 602. If yes, the candidate summary information 602 can be determined as the target summary information 603. If no, the processing of the candidate summary information 602 can be temporarily suspended. After obtaining multiple target summary information 603, the object type 604 can be determined according to the multiple target summary information 603.

[0105] For each file 601, a file name 606 may be generated according to the object type 604 and the candidate summary information 602. After the file name 606 is obtained, directory information 608 may be generated according to the file names 606 of the plurality of files 601 and the to-be-filled directory 607 generated based on the template for generating the directory, and a second database 612 for storing information related to the file 601 may be updated.

[0106] In one example, the information related to the file 601 may include at least one of the following: object identifier, file type, file category, update time, creation time, file name, original file name, candidate summary information, file text information, and file status, etc. The file type refers to the directory where the file 601 is stored, and the file category refers to the multimodal type of the file 601, such as plain text, audio, picture, and mixed. The file status refers to whether the file 601 is valid or whether the file 601 is deleted.

[0107] For the target object, an object name 605 may be generated according to the object type 604 and the plurality of target summary information 603. After obtaining the object name 605, summary information 610 may be generated according to the plurality of target summary information 603, the directory information 608, and the summary to be filled 609 generated based on the template for generating the summary, and a first database 611 for storing information related to the target object may be updated.

[0108] In one example, the information related to the target object may include at least one of the following: object identification, object name, object type, update time, creation time, summary information, object status, etc. The object status refers to whether the target object is valid or whether the target object is deleted. It should be noted that the first database 611 and the second database 612 can be associated through the relationship between the target object and the file, for example, through the object identification.

[0109] According to an embodiment of the present disclosure, by utilizing a first database to store information related to a target object, utilizing a second database to store information related to a file, and associating the two databases through the relationship between the target object and the file, the efficiency of data management as well as the efficiency of query and update are improved, thereby improving the scalability and practicality in complex data scenarios.

[0110] The above are only exemplary embodiments, but are not limited thereto, and may also include other large model-based library update methods known in the art, as long as the efficiency and accuracy of library update can be improved.

[0111] The above describes the library update method based on the large model. Figure 7 A schematic description of the question-answering process based on the large model is given.

[0112] Figure 7 The flowchart of the big model-based question-answering method according to an embodiment of the present disclosure is schematically shown.

[0113] like Figure 7 As shown, the large model-based question-answering method 700 includes operations S710 to S730.

[0114] In operation S710, in response to receiving a question and answer request, a target library to be matched is determined according to the intention of the question text in the question and answer request.

[0115] In operation S770, the target library is matched according to the question text to obtain a matching result.

[0116] In operation S730, the answer text is determined based on the matching result using the large model.

[0117] After receiving the question-and-answer request, the intent of the question text in the question-and-answer request may be determined. For example, the question text may be semantically analyzed to obtain the intent. The intent may be used to characterize the scope of matching the question text.

[0118] After obtaining the intent of the question text, the target library to be matched can be determined according to the intent of the question text. The intent can include one of the following: single database matching or joint database matching. For example, in the case where the intent represents a single database matching, it can be determined that the target library to be matched includes the first database. Alternatively, in the case where the intent represents a joint database matching, it can be determined that the target library to be matched includes the first database and the second database. The target library can be obtained using a library update method based on a large model.

[0119] After determining the target library to be matched, the target library can be matched according to the question text to obtain a matching result. The method of matching the target library can be configured according to actual business needs and is not limited here. For example, the method of matching the target library can include at least one of the following: semantic matching and keyword matching. In the case where the target library includes multiple libraries, the outputs of each library can be combined, deduplication and classification can be performed to obtain the final matching result.

[0120] After obtaining the matching results, the matching results can be sorted according to relevance and importance, and the results that best match the question text can be displayed first. On this basis, the matching results can be enhanced based on the historical records of the user's interaction with the big model to obtain the answer text. The historical records refer to the historical conversations between the user and the big model.

[0121] According to the embodiments of the present disclosure, by determining the target library of the private domain based on the intent analysis of the question text, searching and accurately matching the target library according to the question text, and then using the big model to generate the answer text according to the matching results, the intelligence level of the question-answering process is improved and the accuracy of the answer text is guaranteed.

[0122] The above describes the question-answering method based on the big model. Figure 8 The question-answering process based on the large model is further schematically described.

[0123] Figure 8 An example schematic diagram of a large model-based question-answering process according to an embodiment of the present disclosure is schematically shown.

[0124] like Figure 8 As shown in 800, a user can enter a question text in an input box 802 of a front-end interface 801 and click a search control 803 to initiate a question-and-answer request. The front-end interface 801 can send the question-and-answer request to a large model at the back end so that the large model can perform a question-and-answer process based on the question-and-answer request.

[0125] After receiving the Q&A request, the big model can determine the intent of the question text in the Q&A request. The intent can include single database matching or joint database matching. Single database matching can refer to querying and data matching operations in a separate database. Joint database matching can refer to querying and data matching operations in multiple databases.

[0126] In one example, where a single database match is intended, it may be determined that the target repository includes a first database for storing information related to the target object.

[0127] In another example, when the intention is joint database matching, it can be determined that the target library includes a first database for storing information related to the target object and a second database for storing information related to the file. The first database and the second database can be associated through the relationship between the target object and the file.

[0128] According to the embodiments of the present disclosure, by dynamically determining single-library matching or joint-library matching according to user intent, the flexibility, efficiency and accuracy in processing complex queries are improved, providing users with a more efficient and intelligent query experience.

[0129] The target library may include multiple information to be matched. Each information to be matched corresponds to a vector index and multiple scalar indexes. The vector index may be determined based on the feature vector of the information to be matched, and is used to characterize the content, semantics and other features of the information to be matched, and is suitable for semantic search or similarity retrieval. The scalar index may be determined based on the fields in the information to be matched, and is used to quickly retrieve numerical fields (such as ID, date, etc.) or categorical fields (such as tags, categories, etc.).

[0130] In one example, semantic matching can be performed on multiple vector indexes according to the feature vector of the question text to obtain a first matching result. For example, the feature vector and the multiple vector indexes can be matched based on vector similarity respectively. The first matching result can include information to be matched corresponding to the vector index when the feature vector and the vector index of the question text meet the first predetermined condition.

[0131] The first predetermined condition can be configured according to actual business needs and is not limited here. For example, the first predetermined condition can be that the first similarity between the feature vector and the vector index is greater than the first similarity threshold. Alternatively, the first predetermined condition can also be to limit the number and ranking of the returned matching results.

[0132] In another example, keyword matching can be performed on multiple scalar indexes according to the key fields in the question text to obtain a second matching result. The second matching result may include information to be matched corresponding to the scalar index when the key fields in the question text and the scalar index meet the second predetermined condition.

[0133] The second predetermined condition can be configured according to actual business needs and is not limited here. For example, the second predetermined condition can be that the second similarity between the key field and the scalar index is greater than the second similarity threshold. Alternatively, the second predetermined condition can also be to limit the number and ranking of the returned matching results.

[0134] After obtaining the first matching result and the second matching result, the big model can integrate the first matching result and the second matching result, and combine the historical conversation between the user and the big model to generate an answer text, and display the answer text to the user in the display box 804 of the front-end interface 801.

[0135] According to an embodiment of the present disclosure, by combining a dual matching mechanism of a semantic matching method based on a vector index and a keyword matching method based on a scalar index, the query processing capability, as well as the intelligence level and accuracy of matching are improved.

[0136] The above are only exemplary embodiments, but are not limited thereto and may also include other large model-based question-answering methods known in the art, as long as they can improve the intelligence level of the question-answering process and ensure the accuracy of the answer text.

[0137] The above describes the question-answering method based on the big model. Fig. 9 A schematic description of the resource sharing process based on the large model is given.

[0138] Fig. 9 The flowchart of the resource sharing method based on the large model according to the embodiment of the present disclosure is schematically shown.

[0139] like Fig. 9 As shown, the large model-based resource sharing method 900 includes operations S910 to S920.

[0140] In operation S910 , in response to receiving a sharing request, summary information of the target object is processed according to the target object selected in the library indicated by the sharing request to obtain processed information.

[0141] In operation S920, resource sharing is performed on the processed information.

[0142] The sharing request is used to indicate the target object selected in the library to be shared. The library can be obtained using a library update method based on a large model. After receiving the sharing request, the summary information corresponding to the target object can be extracted from the library, and the summary information can be processed to obtain content that conforms to the story script format. The processing method can be configured according to actual business needs and is not limited here. For example, a natural language generation algorithm can be used to convert the summary information into a coherent text paragraph to obtain processed information. Alternatively, the summary information can be rearranged to conform to the narrative logic of the story to obtain processed information. Alternatively, sensitive words in the summary information can be disturbed to obtain processed information.

[0143] After the processed information is obtained, the processed information can be shared as a resource. The resource sharing method can be configured according to actual business needs and is not limited here. For example, the resource sharing method may include at least one of the following: a method based on a file sharing system, a method based on an API interface, a method based on a database, a method based on a social media or content platform, a method based on access permission control, and a method based on version control, etc.

[0144] According to the embodiments of the present disclosure, by responding to sharing requests, processing summary information of the target object to generate processed information and performing resource sharing, the needs of resource sharing and cross-platform delivery are met, and the automation level, accuracy and efficiency of resource sharing are improved.

[0145] In one example, the summary information may include multiple target fields and field values ​​of each of the multiple target fields. The target field refers to the field that you want to replace or blur, such as certain key fields in the text. For example, the target field may include at least one of the following: name, time, and place, etc. After obtaining the summary information, for each target field, the field value corresponding to the target field can be located according to the target field, and the field value of the target field can be replaced according to the preset value of the target field to obtain the replaced information. The preset value refers to a pre-defined value for replacement, which is usually standardized data without specific reference information or an expected field value.

[0146] After obtaining the replaced information, the replaced information can be disturbed using the large model to obtain processed information. Disturbance processing refers to disturbing the meaningless information in the main line of information so that there are differences on the basis of maintaining the original information. The method of disturbance processing can be configured according to actual business needs and is not limited here.

[0147] For example, the perturbation processing method may include at least one of the following: semantic perturbation, structural perturbation, and detail perturbation. Semantic perturbation refers to changing semantically similar words or phrases to make the replaced information subtly changed while retaining the original meaning. Structural perturbation refers to changing the structure of sentences or the order of paragraphs to affect the fluidity of the replaced information, thereby creating a new narrative style. Detail perturbation refers to adding or reducing details while keeping the main line of the processed information unchanged, so that the generated processed information is more varied in narrative.

[0148] According to an embodiment of the present disclosure, by replacing the field value of the target field in the summary information and combining the disturbance processing of the replaced information with a large model, the privacy protection capability during resource sharing and the diversity of the generated processed information are improved.

[0149] The above describes the resource sharing method based on the large model. Fig.10 The resource sharing process based on the large model is further schematically described.

[0150] Fig.10 An example schematic diagram of a resource sharing process according to an embodiment of the present disclosure is schematically shown.

[0151] like Fig.10 As shown in 1000, the display area 1002 of the front-end interface 1001 can display the user's own information. The recommended attention area 1003 of the front-end interface 1001 can display other user information that the user is recommended to pay attention to.

[0152] After obtaining the processed information, the user can drag the processed information to the input area 1004 of the front-end interface 1001 to upload the processed information. In addition, the front-end interface 1001 can also include a peer exchange community area 1005. The target community displayed in the peer exchange community area 1005 can be determined based on at least one of the user's geographical information and the object type of the target object. For example, target community 1 is determined based on the user's geographical information, target community 2 is determined based on the object type of the target object, and target community 3 is determined based on the user's geographical information and the object type of the target object. The user can share the processed information to any target community by clicking on the front-end interface 1001.

[0153] According to the embodiments of the present disclosure, by determining the target community based on the user's regional information and the object type of the target object, the accuracy of resource sharing is improved, which is conducive to the targeted promotion of resources.

[0154] The above are only exemplary embodiments, but are not limited thereto, and may also include other large model-based resource sharing methods known in the art, as long as the accuracy of resource sharing can be improved.

[0155] The above describes the resource sharing method based on the large model. Fig.11 The information display process is schematically described.

[0156] Fig.11 The flowchart of the information display method according to the embodiment of the present disclosure is schematically shown.

[0157] like Fig.11 As shown, the information display method 1100 includes operations S1110~S1120.

[0158] In operation S1110 , in response to receiving a user request, a pending request is sent.

[0159] In operation S1120, in response to obtaining the response information generated in response to the pending request, the response information is displayed.

[0160] In the case where the user request is a file upload operation, the pending request may be an update request, and the front end may send the update request to the big model at the back end. After receiving the update request, the big model may respond to the update request using a library update method based on the big model to obtain object information, and the back end may send the object information to the front end. Thus, the front end may display the object information generated in response to the update request to implement the update of the library.

[0161] In the case where the user request is a question-and-answer operation, the request to be processed may be a question-and-answer request, and the front end may send the question-and-answer request to the big model at the back end. After receiving the question-and-answer request, the big model may respond to the question-and-answer request using a question-and-answer method based on the big model to obtain an answer text, and the back end may send the answer text to the front end. Thus, the front end may display the answer text generated in response to the question-and-answer request to implement question-and-answer.

[0162] In the case where the user request is a sharing operation, the request to be processed may be a sharing request, and the front end may send the sharing request to the big model at the back end. After receiving the sharing request, the big model may respond to the sharing request using a resource sharing method based on the big model to obtain processed information, and the back end may send the processed information to the front end. Thus, the front end may display the processed information generated in response to the sharing request to achieve resource sharing.

[0163] According to the embodiments of the present disclosure, by responding to user requests and displaying response information generated based on different methods, this multi-functional processing capability, real-time response mechanism and automated display function provide users with a convenient and efficient operating experience and improve the flexibility and intelligence of information display.

[0164] The above are only exemplary embodiments, but are not limited thereto, and may also include other information display methods known in the art, as long as the flexibility and intelligence of information display can be improved.

[0165] It should be noted that in the embodiments of the present invention, certain software, components, models and other existing solutions in the industry may be mentioned, which should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present invention, but it does not mean that the present invention has or must use this solution.

[0166] Based on the library update method based on the large model provided by the present disclosure, the present disclosure also provides a library update device based on the large model. Fig.12 The device is described in detail.

[0167] Fig.12 A block diagram of a library updating device based on a large model according to an embodiment of the present disclosure is schematically shown.

[0168] like Fig.12 As shown, the large model-based library updating apparatus 1200 may include a first determining module 1210 , a first processing module 1220 and an updating module 1230 .

[0169] The first determination module 1210 is configured to, in response to receiving an update request, determine an object type of the target object according to candidate summary information of each of the multiple files of the target object included in the update request, wherein the multiple files have multiple file types.

[0170] The first processing module 1220 is used to process a plurality of candidate summary information according to a preset template for an object type to obtain object information.

[0171] The updating module 1230 is used to update the library according to the object information, the multiple files and the candidate summary information of each of the multiple files.

[0172] According to an embodiment of the present disclosure, the first determining module 1210 may include a first determining unit and a second determining unit.

[0173] A first determining unit is configured to determine, from among the plurality of candidate summary information, a plurality of target summary information belonging to a target type according to the intention of each candidate summary information; and

[0174] The second determination unit is used to determine the object type according to the multiple target summary information by using the large model.

[0175] According to an embodiment of the present disclosure, the large model-based library updating apparatus 1200 may further include a first generating module and a second generating module.

[0176] The first generating module is used to generate an object name for a target object according to the object type and a plurality of target summary information.

[0177] The second generating module is used to generate a file name for each file according to the object type and the candidate summary information, so as to obtain the respective file names of the plurality of files.

[0178] According to an embodiment of the present disclosure, the preset template includes a template for generating a directory and a template for generating a summary, and the object information includes directory information and summary information for a target object.

[0179] According to an embodiment of the present disclosure, the first processing module 1220 may include a first generating unit and a second generating unit.

[0180] The first generating unit is used to generate directory information by using the large model according to the file names of the plurality of files and the directory to be filled generated based on the template for generating the directory.

[0181] The second generating unit is used to generate summary information by using the large model according to the multiple target summary information, the directory information, and the summary to be filled generated based on the template for generating the summary.

[0182] According to an embodiment of the present disclosure, the library includes a first database and a second database, the first database is used to store information related to the target object, the second database is used to store information related to the file, and the first database and the second database are associated through the relationship between the target object and the file.

[0183] According to an embodiment of the present disclosure, the update module 1230 may include a first update unit and a second update unit.

[0184] The first updating unit is used to update the first database according to the object identifier, object type, object name and summary information of the target object.

[0185] The second updating unit is used to update the second database according to the file name, file type, candidate summary information and directory information of each file.

[0186] According to an embodiment of the present disclosure, candidate summary information may be determined in the following manner.

[0187] Using the large model, a summary generation tool is called based on the file type. Based on the summary generation tool, the file is processed to obtain candidate summary information.

[0188] According to an embodiment of the present disclosure, the target objects include legal cases in the legal field.

[0189] Based on the large model-based question-answering method provided by the present disclosure, the present disclosure also provides a large model-based question-answering device. Fig.13 The device is described in detail.

[0190] Fig.13 A block diagram of a question-answering device based on a large model according to an embodiment of the present disclosure is schematically shown.

[0191] like Fig.13 As shown, the large model-based question-answering device 1300 may include a second determination module 1310 , a matching module 1320 , and a third determination module 1330 .

[0192] The second determination module 1310 is used to determine a target library to be matched according to the intention of the question text in the question and answer request in response to receiving the question and answer request.

[0193] The matching module 1320 is used to match the target library according to the question text to obtain a matching result, wherein the target library is obtained by updating the library based on the large model.

[0194] The third determination module 1330 is used to determine the answer text based on the matching results using the large model.

[0195] According to an embodiment of the present disclosure, the target library includes multiple information to be matched, each information to be matched corresponds to a vector index and multiple scalar indexes, the vector index is determined according to the feature vector of the information to be matched, and the scalar index is determined according to the field in the information to be matched.

[0196] According to an embodiment of the present disclosure, the matching module 1320 may include a first matching unit and a second matching unit.

[0197] The first matching unit is used to perform semantic matching on multiple vector indexes according to the feature vector of the question text to obtain a first matching result, wherein the first matching result includes information to be matched corresponding to the vector index when the feature vector of the question text and the vector index meet a first predetermined condition.

[0198] The second matching unit is used to perform keyword matching on multiple scalar indexes according to the key fields in the question text to obtain a second matching result, wherein the second matching result includes the information to be matched corresponding to the scalar index when the key fields and the scalar index in the question text meet the second predetermined condition.

[0199] According to an embodiment of the present disclosure, the second determining module 1310 may include a third determining unit and a fourth determining unit.

[0200] The third determining unit is used to determine, when the intention is a single database match, that the target database includes a first database, wherein the first database is used to store information related to the target object.

[0201] The fourth determination unit is used to determine, when the intention is joint database matching, that the target database includes a first database and a second database, wherein the second database is used to store information related to the file, and the first database and the second database are associated through the relationship between the target object and the file.

[0202] Based on the resource sharing method based on the big model provided by the present disclosure, the present disclosure also provides a resource sharing device based on the big model. Fig.14 The device is described in detail.

[0203] Fig.14 A block diagram of a resource sharing device based on a large model according to an embodiment of the present disclosure is schematically shown.

[0204] like Fig.14 As shown, the large model-based resource sharing apparatus 1400 may include a second processing module 1410 and a resource sharing module 1420 .

[0205] The second processing module 1410 is used to process the summary information of the target object in response to receiving the sharing request according to the selected target object in the library indicated by the sharing request to obtain processed information, wherein the library is obtained by using the library update based on the large model.

[0206] The resource sharing module 1420 is used to share the processed information.

[0207] According to an embodiment of the present disclosure, the summary information includes a plurality of target fields and respective field values ​​of the plurality of target fields.

[0208] According to an embodiment of the present disclosure, the second processing module 1410 may include a replacement unit and a disturbance processing unit.

[0209] The replacement unit is used to replace the field value of each target field according to the preset value of the target field to obtain replaced information.

[0210] A disturbance processing unit is used to perturb the replaced information using a large model

[0211] According to an embodiment of the present disclosure, the resource sharing module 1420 may include a fifth determining unit and a sharing unit.

[0212] The fifth determining unit is configured to determine the target community according to at least one of the region information of the user who selects the target object and the object type of the target object.

[0213] The sharing unit is used to share the processed information to the target community.

[0214] Based on the information display method provided by the present disclosure, the present disclosure also provides an information display device. Fig.15 The device is described in detail.

[0215] Fig.15 A block diagram of an information display device according to an embodiment of the present disclosure is schematically shown.

[0216] like Fig.15 As shown, the information display device 1500 may include a sending module 1510 and a display module 1520 .

[0217] The sending module 1510 is used to send a pending request in response to receiving a user request, wherein the pending request includes at least one of an update request, a question-and-answer request, and a sharing request.

[0218] Display module 1520 is used to display response information in response to obtaining response information generated in response to the pending request, wherein the response information is generated using a big model-based library update, a big model-based question and answer, or a big model-based resource sharing, and the response information includes at least one of object information, answer text, and processed information.

[0219] Fig.16 The structural block diagram of the intelligent agent of the large model according to the embodiment of the present disclosure is schematically shown.

[0220] In the embodiments of the present disclosure, inspired by the von Neumann structure in modern computer theory, such as Fig.16 As shown, the AI ​​agent 1600 may include five core modules: an input module 1610 , a control module 1620 , a storage module 1630 , a calculation module 1640 and an output module 1650 .

[0221] The input module 1610 is responsible for receiving or sensing information such as queries, requests, instructions, signals or data from the outside world (such as users or external environment), and converting them into a format that can be understood and processed by the AI ​​agent 1600. The input module 1610 is the primary link for the AI ​​agent 1600 to interact with the outside world. It enables the AI ​​agent 1600 to efficiently and accurately obtain the necessary "sensory" information from the outside world and respond to this information.

[0222] In an example, the input module 1610 may input the update request, question and answer request, and sharing request described above.

[0223] The control module 1620 is the core support for the AI ​​agent 1600 to handle complex tasks. The control module 1620 can execute the library update method based on the large model, the question-answering method based on the large model, and the resource sharing method based on the large model described above.

[0224] In the example, the control module 1620 will continuously interact with the storage module 1630, the operation module 1640 and / or the output module 1650 during operation. However, it should be noted that in the embodiment of the present disclosure, the control module 1620 acts as a single initiator to initiate communication with the storage module 1630, the operation module 1640 and / or the output module 1650, and there is no communication coupling between the storage module 1630, the operation module 1640 and the output module 1650.

[0225] In the example, the performance of the control module 1620 may be closely related to the large model on which the AI ​​agent 1600 is based. In order to give full play to the capabilities of the large language model, the internal structure of the control module 1620 may be designed to be highly configurable and extensible to cope with various types of tasks and requirements in real scenarios.

[0226] The storage module 1630 may be responsible for memorizing information such as historical conversations, event flows, etc. The preset templates, the first database, and the second database as described above may be included in the storage module 1630 .

[0227] In the example, after the AI ​​agent 1600 obtains the update request, the AI ​​agent 1600 may determine the object type of the target object. The object type may be stored in the storage module 1630. The AI ​​agent 1600 may retrieve the preset template for the object type from the storage module 1630 and feed it back to the control module 1620. Then, the control module 1620 may obtain the object information using the preset template for the object type fed back, and pass the object information to the output module 1650.

[0228] The operation module 1640 can be regarded as a predefined tool library. Information perturbation tools for perturbing the replaced information can be included in the operation module 1640.

[0229] In the example, when the AI ​​agent 1600 needs to perform disturbance processing on the replaced information, the relevant information disturbance tool can be called from the operation module 1640 and fed back to the control module 1620. Then, the control module 1620 can use the information disturbance tool fed back to perform disturbance processing on the replaced information, obtain processed information, and pass the processed information to the output module 1650. It can be understood that although the large language model has excellent language understanding and generation capabilities, it is the same as a human being. Without the help of any tools, the tasks that can be solved are very limited. When the AI ​​agent 1600 is given the ability to call tools, it can achieve tasks such as completing summary generation with the help of summary generation tools and completing information disturbance with the help of information disturbance tools.

[0230] The output module 1650 may output the object information, answer text, and processed information described above.

[0231] The AI ​​agent 1600 according to the embodiment of the present disclosure can simply and effectively improve the level of intelligence, and improve flexibility and versatility.

[0232] Fig.17 A block diagram of an electronic device suitable for implementing library updates, question and answer, resource sharing methods, and information display methods based on a large model according to an embodiment of the present disclosure is schematically shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0233] like Fig.17 As shown, the device 1700 includes a computing unit 1701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1702 or a computer program loaded from a storage unit 1708 into a random access memory (RAM) 1703. In the RAM 1703, various programs and data required for the operation of the device 1700 can also be stored. The computing unit 1701, the ROM 1702, and the RAM 1703 are connected to each other via a bus 1704. An input / output (I / O) interface 1705 is also connected to the bus 1704.

[0234] A number of components in the device 1700 are connected to the I / O interface 1705, including: an input unit 1706, such as a keyboard, a mouse, etc.; an output unit 1707, such as various types of displays, speakers, etc.; a storage unit 1708, such as a disk, an optical disk, etc.; and a communication unit 1709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1709 allows the device 1700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0235] The computing unit 1701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 1701 performs the various methods and processes described above, such as library updates, question and answer, resource sharing methods, and information display methods based on large models. For example, in some embodiments, library updates, question and answer, resource sharing methods, and information display methods based on large models may be implemented as computer software programs, which are tangibly included in machine-readable media, such as storage unit 1708. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 1700 via ROM 1702 and / or communication unit 1709. When the computer program is loaded into RAM 1703 and executed by the computing unit 1701, one or more steps of the library update, question and answer, resource sharing method, and information display method based on the large model described above may be performed. Alternatively, in other embodiments, the computing unit 1701 may be configured to perform the library update, question and answer, resource sharing method, and information display method based on the large model in any other appropriate manner (e.g., by means of firmware).

[0236] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0237] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0238] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0239] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0240] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0241] A computer system may include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises through computer programs running on respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server combined with a blockchain.

[0242] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0243] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A library update method based on a large model, comprising: In response to receiving an update request, determining, for a plurality of files of a target object included in the update request, an object type of the target object according to candidate summary information of each of the plurality of files, wherein the plurality of files have a plurality of file types; Processing the plurality of candidate summary information according to a preset template for the object type to obtain object information; and A library is updated based on the object information, the plurality of files, and candidate summary information of each of the plurality of files.

2. The method according to claim 1, wherein: The determining the object type of the target object according to the candidate summary information of each of the multiple files includes: According to the intention of each of the candidate summary information, determining a plurality of target summary information belonging to a target type from among the plurality of candidate summary information; and The object type is determined by using the large model and based on the plurality of target summary information.

3. The method according to claim 2, further comprising, before processing the plurality of candidate summary information according to the preset template of the object type to obtain the object information: For the target object, generating an object name according to the object type and the plurality of target summary information; and For each of the files, a file name is generated according to the object type and the candidate summary information, so as to obtain the respective file names of the multiple files.

4. The method according to claim 2, wherein: The preset templates include a template for generating a directory and a template for generating a summary, and the object information includes directory information and summary information for the target object; The processing of the plurality of candidate summary information according to the preset template of the object type to obtain the object information includes: Generate the directory information by using the large model according to the file names of the plurality of files and the directory to be filled generated based on the template for generating the directory; as well as The summary information is generated by using the large model according to the plurality of target summary information, the directory information, and the summary to be filled generated based on the template for generating a summary.

5. The method according to claim 4, wherein: The library includes a first database and a second database, the first database is used to store information related to the target object, the second database is used to store information related to the file, and the first database and the second database are associated through the relationship between the target object and the file; The updating of the library according to the object information, the plurality of files and the candidate summary information of each of the plurality of files comprises: For the target object, updating the first database according to the object identifier of the target object, the object type, the object name and the summary information; as well as For each of the files, the second database is updated according to the file name, the file type, the candidate summary information and the directory information.

6. The method according to any one of claims 1 to 5, wherein: The candidate summary information is determined in the following manner: Using the large model, calling a summary generation tool determined based on the file type of the file; and Based on the summary generation tool, the file is processed to obtain the candidate summary information.

7. The method according to any one of claims 1 to 5, wherein: The target objects include legal cases in the legal field.

8. A question answering method based on a large model, comprising: In response to receiving a question-and-answer request, determining a target library to be matched according to the intent of the question text in the question-and-answer request; According to the question text, the target library is matched to obtain a matching result, wherein the target library is obtained by using the method according to any one of claims 1 to 7; as well as The answer text is determined based on the matching result using the large model.

9. The method according to claim 8, wherein: The target library includes a plurality of information to be matched, each of which corresponds to a vector index and a plurality of scalar indexes, wherein the vector index is determined according to a feature vector of the information to be matched, and the scalar index is determined according to a field in the information to be matched; According to the question text, the target library is matched to obtain a matching result, which includes: According to the feature vector of the question text, semantic matching is performed on the plurality of vector indexes to obtain a first matching result, wherein the first matching result includes the to-be-matched information corresponding to the vector index when the feature vector of the question text and the vector index satisfy a first predetermined condition; and According to the key fields in the question text, keyword matching is performed on the multiple scalar indexes to obtain a second matching result, wherein the second matching result includes the information to be matched corresponding to the scalar index when the key fields in the question text and the scalar index meet a second predetermined condition.

10. The method according to claim 8 or 9, wherein: Determining the target library to be matched according to the intention of the question text in the question-answer request includes: In the case where the intention is a single database match, determining that the target database includes a first database, wherein the first database is used to store information related to the target object; and In the case where the intention is joint database matching, it is determined that the target database includes the first database and a second database, wherein the second database is used to store information related to the file, and the first database and the second database are associated through the relationship between the target object and the file.

11. A resource sharing method based on a large model, comprising: In response to receiving a sharing request, processing summary information of the target object according to a selected target object in a library indicated by the sharing request to obtain processed information, wherein the library is obtained by using the method of any one of claims 1 to 7; and The processed information is resource shared.

12. The method according to claim 11, wherein: The summary information includes a plurality of target fields and respective field values ​​of the plurality of target fields; The summary information of the target object is processed to obtain the processed information, including: For each of the target fields, the field value of the target field is replaced according to the preset value of the target field to obtain replaced information; as well as The replaced information is perturbed using the large model to obtain the processed information.

13. The method according to claim 11 or 12, wherein: The resource sharing of the processed information includes: determining a target community according to at least one of the region information of the user who selected the target object and the object type of the target object; and The processed information is shared with the target community.

14. A method for displaying information, comprising: In response to receiving the user request, sending a pending request, wherein the pending request includes at least one of an update request, a question-and-answer request, and a sharing request; and In response to obtaining response information generated in response to the pending request, displaying the response information, wherein the response information is generated using the method described in any one of claims 1 to 13, and the response information includes at least one of object information, answer text, and processed information.

15. A library updating device based on a large model, comprising: A first determination module is configured to, in response to receiving an update request, determine, for a plurality of files of a target object included in the update request, an object type of the target object according to candidate summary information of each of the plurality of files, wherein the plurality of files have a plurality of file types; A first processing module, configured to process the plurality of candidate summary information according to a preset template for the object type to obtain object information; and An updating module is used to update a library according to the object information, the multiple files and the candidate summary information of each of the multiple files.

16. The device according to claim 15, wherein: The first determining module comprises: A first determining unit is configured to determine, from among the plurality of candidate summary information, a plurality of target summary information belonging to a target type according to the intention of each of the candidate summary information; and The second determination unit is used to determine the object type according to the plurality of target summary information by using the large model.

17. The apparatus according to claim 16, further comprising: A first generating module, configured to generate an object name for the target object according to the object type and the plurality of target summary information; as well as The second generating module is used to generate a file name for each of the files according to the object type and the candidate summary information, so as to obtain the file names of the respective files.

18. The device according to claim 16, wherein: The preset templates include a template for generating a directory and a template for generating a summary, and the object information includes directory information and summary information for the target object; The first processing module comprises: A first generating unit, configured to generate the directory information by using the large model according to the file names of the plurality of files and the directory to be filled generated based on the template for generating the directory; as well as The second generating unit is used to generate the summary information by using the large model according to the plurality of target summary information, the directory information, and the summary to be filled generated based on the template for generating the summary.

19. The device according to claim 18, wherein: The library includes a first database and a second database, the first database is used to store information related to the target object, the second database is used to store information related to the file, and the first database and the second database are associated through the relationship between the target object and the file; The update module includes: A first updating unit, configured to update the first database according to the object identifier, the object type, the object name and the summary information of the target object; as well as The second updating unit is used to update the second database for each of the files according to the file name, the file type, the candidate summary information and the directory information.

20. The device according to any one of claims 15 to 19, wherein: The candidate summary information is determined in the following manner: Using the large model, calling a summary generation tool determined based on the file type of the file; and Based on the summary generation tool, the file is processed to obtain the candidate summary information.

21. The device according to any one of claims 15 to 19, wherein: The target objects include legal cases in the legal field.

22. A question-answering device based on a large model, comprising: A second determination module is used to determine, in response to receiving a question and answer request, a target library to be matched according to the intention of the question text in the question and answer request; A matching module, used for matching the target library according to the question text to obtain a matching result, wherein the target library is obtained by using the device according to any one of claims 15 to 21; as well as The third determination module is used to determine the answer text according to the matching result using the large model.

23. The device according to claim 22, wherein: The target library includes a plurality of information to be matched, each of which corresponds to a vector index and a plurality of scalar indexes, wherein the vector index is determined according to a feature vector of the information to be matched, and the scalar index is determined according to a field in the information to be matched; The matching module comprises: A first matching unit, configured to perform semantic matching on the plurality of vector indexes according to the feature vector of the question text to obtain a first matching result, wherein the first matching result includes the to-be-matched information corresponding to the vector index when the feature vector of the question text and the vector index satisfy a first predetermined condition; and The second matching unit is used to perform keyword matching on the multiple scalar indexes according to the key fields in the question text to obtain a second matching result, wherein the second matching result includes the information to be matched corresponding to the scalar index when the key fields in the question text and the scalar index meet a second predetermined condition.

24. The device according to claim 22 or 23, wherein: The second determining module comprises: A third determining unit is configured to determine, when the intention is a single database match, that the target database includes a first database, wherein the first database is used to store information related to the target object; and The fourth determination unit is used to determine, when the intention is a joint database match, that the target library includes the first database and the second database, wherein the second database is used to store information related to the file, and the first database and the second database are associated through the relationship between the target object and the file.

25. A resource sharing device based on a large model, comprising: a second processing module, configured to, in response to receiving a sharing request, process summary information of the target object according to a selected target object in a library indicated by the sharing request to obtain processed information, wherein the library is obtained by using the apparatus of any one of claims 15 to 21; and The resource sharing module is used to share the processed information.

26. The device according to claim 25, wherein The summary information includes a plurality of target fields and respective field values ​​of the plurality of target fields; The second processing module comprises: a replacing unit, configured to replace, for each of the target fields, a field value of the target field according to a preset value of the target field, to obtain replaced information; and A disturbance processing unit is used to perform disturbance processing on the replaced information using the large model to obtain the processed information.

27. The device according to claim 25 or 26, wherein: The resource sharing module includes: A fifth determining unit, configured to determine a target community according to at least one of the region information of a user who selects the target object and the object type of the target object; and A sharing unit is used to share the processed information with the target community.

28. An information display device, comprising: a sending module, configured to send a pending request in response to receiving a user request, wherein the pending request includes at least one of an update request, a question-and-answer request, and a sharing request; and A display module, used for displaying the response information in response to obtaining the response information generated in response to the pending request, wherein the response information is generated using the device described in any one of claims 15 to 27, and the response information includes at least one of object information, answer text and processed information.

29. An agent configured to execute the method according to any one of claims 1 to 14.

30. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 14.

31. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 14 are implemented.

32. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 14 are implemented.

Images