Question and answer method and device based on personalized data of user

Abstract

The invention provides a question and answer method and device based on personalized data of a user. The method comprises the steps that a data backflow system obtains the personalized data of the user; processing the personalized data, and filling the personalized data into the defined structured pattern and the unstructured personalized information of the personalized schema, wherein the structured pattern and the unstructured personalized information are defined by the personalized schema; the personalized data processing system is used for obtaining structured schema data defined by personalized schema and unstructured personalized data described by a natural language through AI processing; the online personalized content generation module is used for matching a corresponding prompt word template in a prompt warehouse; and inputting a personalized cue word corresponding to the user into the large model to generate a corresponding personalized reply content. According to the method, on the basis of an LLM large model technology, user personalized data obtained through an offline personalized data processing flow is input, and the cue word template of multi-scheme combination is generated in combination with triggering, so that the purpose of generating corresponding appropriate solutions for different users can be achieved.

Description

Technical Field

[0001] The invention relates to a question-answering method and a device based on personalized data of a user. Background Art

[0002] In the service form of traditional question-and-answer systems, the answers granted to different users are homogeneous, and it is impossible to provide users with accurate answer solutions for each user. Summary of the invention

[0003] The object of the present invention is to provide a question-answering method and device based on user's personalized data.

[0004] To solve the above problems, the present invention provides a question-answering method based on user's personalized data, comprising:

[0005] The data return system obtains the user's personalized data;

[0006] The data return system processes the personalized data and fills it into the structured pattern and unstructured personalized information defined in the personalized schema;

[0007] The personalized data processing system, based on the personalized data filled into the structured pattern defined by the personalized schema and the unstructured personalized information, obtains the structured schema data defined by the personalized schema and the unstructured personalized data described in natural language through AI processing;

[0008] Personalized schema management system, which stores structured schema data corresponding to each user and unstructured personalized data described in natural language;

[0009] The online personalized content generation module obtains the structured schema data corresponding to the user and the unstructured personalized data described in natural language from the personalized schema management system, matches the corresponding prompt word template in the prompt warehouse based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language; assembles the matched prompt word templates based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language to obtain the personalized prompt word corresponding to the user, and inputs the personalized prompt word corresponding to the user into the big model to generate the corresponding personalized reply content.

[0010] Furthermore, in the above method, the personalized data includes: data on user conversations, user behaviors and user attributes.

[0011] Furthermore, in the above method, the data reflow system processes the personalized data and fills it into the structured pattern and unstructured personalized information defined by the personalized schema, including:

[0012] Extract key information about user intent and user events from the user conversation data, and fill the extracted key information about user intent and user events into the corresponding fields of intent and events in the defined structured pattern of the personalized schema; at the same time, perform relevant processing of summary and description summary from the user conversation data, and fill the summary and description summary into the unstructured personalized information;

[0013] Extract the user usage time and user stay time behavior data from the user behavior data, and fill the extracted relevant data into the usage time and stay time and other related fields in the defined structured pattern of the personalized schema; at the same time, use the LLM technology to generate a natural language description of the user behavior data and fill it into the unstructured personalized information;

[0014] Extract age and gender related fields from user profile data and fill them into the age and gender related fields in the structured pattern defined in the personalized schema; at the same time, use LLM technology to generate a natural language description of the user profile data and fill it into the unstructured personalized information.

[0015] Furthermore, in the above method, the corresponding prompt word template is matched in the prompt warehouse.

[0016] The prompt warehouse includes multiple prompt word templates, each prompt template represents a prompt description of a specific scenario, corresponding structured schema data of different users and unstructured personalized data described in natural language, matching different prompt word templates.

[0017] Furthermore, in the above method, after generating the corresponding personalized reply content, it also includes:

[0018] Based on the user's corresponding structured schema data and unstructured personalized data described in natural language, personalized reply content is displayed in one or more ways: text, pictures, videos, conversations, and live broadcasts.

[0019] Furthermore, in the above method, the personalized prompt words corresponding to the user are input into the large model to generate corresponding personalized reply content, including:

[0020] Input the personalized prompt words corresponding to the user into the large model to generate corresponding personalized reply sub-content;

[0021] Based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language, the personalized reply sub-contents are combined to obtain the corresponding personalized reply content.

[0022] Furthermore, in the above method, after generating the corresponding personalized reply content, it also includes:

[0023] If the personalized reply content is in a conversational manner, the user role customization system defines the user role;

[0024] The AI ​​character customization system is used to define AI characters that communicate with users;

[0025] The AI ​​dialogue system constructs a virtual AI character based on the defined user role and AI role to practice dialogue with the user;

[0026] The AI ​​analysis system analyzes the content of each conversation exercise, points out what the user did well and what needs improvement, and provides timely summaries and feedback to the user.

[0027] Furthermore, in the above method, the personalized prompt words corresponding to the user are input into the large model to generate corresponding personalized reply content, and then the method further includes:

[0028] The AI ​​test generation system generates corresponding personalized test questions based on the corresponding personalized response content;

[0029] The AI ​​test system presents personalized test questions and obtains the user's answers within a limited time.

[0030] The AI ​​test analysis system generates a test analysis report based on the answer content;

[0031] The AI ​​test feedback system adjusts the personalized response content and / or the display method of the personalized response content based on the test analysis report.

[0032] According to another aspect of the present invention, there is further provided a computer-readable storage medium having computer-executable instructions stored thereon, wherein when the computer-executable instructions are executed by a processor, the processor is caused to: execute any of the methods described above.

[0033] According to another aspect of the present invention, there is also provided a computer device, comprising:

[0034] Processor; and

[0035] A memory arranged to store computer executable instructions, which when executed cause the processor to: perform any of the methods described above.

[0036] Compared with the prior art, the intelligent solution system of the present invention is based on the LLM (Large Language Model) technology. The input is the user personalized data obtained through the offline personalized data processing process. Combined with the prompt word template that triggers the generation of multiple solution combinations, it can achieve the purpose of generating corresponding appropriate solutions for different users.

[0037] In addition, in traditional service forms, the way of transmitting information is usually text or video, and the information form is too monotonous. When users use it for a period of time, they often feel bored. During the user's learning process, this application can provide text, pictures, videos, conversations, live broadcasts and other methods to meet the user's learning needs, which can greatly improve the user's learning effect. The entire system of the present invention uses user personalized information, that is, the structured schema data corresponding to the user and the unstructured personalized data described in natural language, as input, and uses the intelligent solution system as the hub. The intelligent solution hub outputs text, pictures, videos, conversations, live broadcasts and other forms to meet the user's needs.

[0038] In addition, this solution innovatively improves the content satisfaction method in traditional service forms, upgrading the unified content method for users in traditional service forms to one based on user personalization. It can disassemble the learning content, reorganize the information structure in the smallest granularity, and recommend the most suitable content to different users. Since the improvement allows different users to see the content that suits them best, it produces higher learning efficiency for users.

[0039] Moreover, in the traditional service form, traditional education methods such as exercises are used to respond to users' practice demands. This method is too passive and not conducive to users' active thinking. Based on AI LLM technology, the present invention can train a variety of role-playing models. Users can also customize their own role-playing models, assist users in practice through dialogue, analyze the dialogue content after each practice, summarize and feedback to users in a timely manner, and ensure the integration of "learning + practice" for users. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of a question-answering method based on user personalized data according to an embodiment of the present invention;

[0041] Figure 2 is a schematic diagram of a display method of personalized content obtained based on personalized data of a user according to an embodiment of the present invention;

[0042] Figure 3is a flow chart of a role-playing practice method according to an embodiment of the present invention;

[0043] Figure 4 is a flow chart of an AI testing method according to an embodiment of the present invention;

[0044] Figure 5 It is a schematic diagram of disassembling learning content according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0046] In a typical configuration of the present application, the terminal, the device of the service network and the trusted party all include one or more processors (CPU), input / output interfaces, network interfaces and memories.

[0047] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0048] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0049] like Figure 1 As shown, the present invention provides a question-answering method based on user's personalized data, the method comprising:

[0050] Step S1, the data return system obtains the user's personalized data;

[0051] Preferably, the personalized data includes: data on user conversations, user behaviors and user attributes;

[0052] Specifically, the data return system can process user conversations, behaviors, profiles and other related data. The data return system can collect relevant data to the big data processing system, where:

[0053] The user conversation data refers to the conversation data collected when the user and the AI ​​conversation service have a conversation;

[0054] The user behavior data refers to the behavior data formed by users on the platform, for example, it may include:

[0055] Conversation, collection, video watching and other behaviors;

[0056] Profile data refers to user attribute information, for example, it may include: gender, age, etc.

[0057] Step S2, the data reflow system processes the personalized data and fills it into the structured pattern and unstructured personalized information defined in the personalized schema;

[0058] Preferably, key information, such as user intent and user events, is extracted from the data of the user conversation, and the extracted user intent, user event and other data are filled into the corresponding relevant fields of intent and event in the structured pattern defined in the personalized schema; at the same time, relevant processing such as summary and description summary is performed on the data of the user conversation, and the results of summary and description summary are filled into the unstructured personalized information;

[0059] Extract user usage time and user stay time from user behavior data, and fill the extracted relevant data into the usage time and stay time fields in the defined structured pattern of the personalized schema; at the same time, use the LLM technology to generate a natural language description of the user behavior data and fill it into the unstructured personalized information;

[0060] Extract relevant fields such as age and gender from user profile data and fill them into relevant fields such as age and gender in the structured pattern defined in the personalized schema; at the same time, use LLM technology to generate a natural language description of the user profile data and fill it into the unstructured personalized information.

[0061] Step S3, the personalized data processing system obtains the structured schema data defined by the personalized schema and the unstructured personalized data described in natural language through AI processing based on the personalized data filled in the structured pattern defined by the personalized schema and the unstructured personalized information;

[0062] Here, the personalized data processing system takes the personalized data filled into the personalized schema and the unstructured personalized information stored in the big data processing system as input, and processes the structured schema data defined in the personalized schema and the unstructured personalized data described in natural language through AI.

[0063] Among them, personalized schema refers to a defined structured pattern, including multiple fields, such as gender, age and other information; unstructured personalized information is described in natural language, such as "I am a programmer who likes reading and traveling." It includes a variety of AI information extraction and processing capabilities, such as summary extraction and intent recognition.

[0064] Step S4: The personalized schema management system stores the structured schema data corresponding to each user and the unstructured personalized data described in natural language.

[0065] Specifically, the personalized schema management system can manage personalized schemas and unstructured personalized information. Each user saves a personalized schema data and a unstructured personalized information in the personalized schema management system.

[0066] Step S5, the online personalized content generation module obtains the structured schema data corresponding to the user and the unstructured personalized data described in natural language from the personalized schema management system, matches the corresponding prompt word template in the prompt warehouse based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language; assembles the matched prompt word templates based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language to obtain the personalized prompt word corresponding to the user, and inputs the personalized prompt word corresponding to the user into the big model to generate the corresponding personalized reply content.

[0067] Here, if Figure 1 As shown in FIG. 1 , the offline personalized data processing process includes three systems: data return system, personalized data processing system, and personalized schema management system. After the three processes are processed, personalized data is obtained.

[0068] The online personalized content generation process uses the user personalized data obtained from the offline personalized data processing process as input, including structured data defined by a personalized schema and unstructured data described in natural language. It matches the appropriate prompt word template in the prompt warehouse and obtains the user personalized prompt word template after assembly. The large model uses the user personalized prompt word as input to generate personalized reply content.

[0069] Specifically, the best prompt word template may be matched in the prompt warehouse based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language.

[0070] Preferably, the prompt warehouse includes a plurality of prompt word templates, each prompt template represents a prompt description of a specific scenario, corresponding structured schema data of different users and unstructured personalized data described in natural language, matching different prompt word templates.

[0071] By combining the structured schema data corresponding to the user, the unstructured personalized data described in natural language, and the prompt word template, a natural language description can be generated. The natural language description is the user personalized prompt word.

[0072] The user's personalized prompt words are used as the input of the big model, and the big model generates the user's personalized reply content.

[0073] Specifically, the big model may include: a general big model and a big model obtained after training based on the general big model. The general big model may use an industry open source model, and the big model obtained after training based on the general big model may be fine-tuned and subjected to reinforcement learning training. General big models of different sizes, such as 7B parameters, 14B parameters, and 70B parameters, may be selected for training.

[0074] The user's personalized prompt words are input as questions and are input into the big model. The big model generates the corresponding output and answer, that is, the user's personalized reply content, which is a natural language text description.

[0075] The intelligent solution system of the present invention is based on the LLM (Large Language Model) technology. The input is the user personalized data obtained through the offline personalized data processing process. Combined with the prompt word template that triggers the generation of multiple solution combinations, it can achieve the purpose of generating corresponding appropriate solutions for different users.

[0076] like Figure 2As shown, in a preferred embodiment, after step S5 generates the corresponding personalized reply content, it also includes:

[0077] Step S6, based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language, the personalized reply content is displayed in one or more ways of text, picture, video, conversation and live broadcast.

[0078] Here, in traditional service forms, the way of transmitting information is usually text or video, and the information form is too monotonous. When users use it for a period of time, they often feel bored. During the user's learning process, this application can provide text, pictures, videos, conversations, live broadcasts and other methods to meet the user's learning needs, which can greatly improve the user's learning effect. The entire system of the present invention uses user personalized information, that is, the structured schema data corresponding to the user and the unstructured personalized data described in natural language, as input, and uses the intelligent solution system as the hub. The intelligent solution hub outputs text, pictures, videos, conversations, live broadcasts and other forms to meet the user's needs.

[0079] Among them, text content is presented to users in the form of articles, etc.

[0080] Audio content is presented to users in the form of music, etc.

[0081] Video content is presented to users in the form of short videos, long videos, etc.

[0082] Image content is presented to users in the form of pictures.

[0083] Conversational content is presented to users in the form of conversations with AI.

[0084] The content in the form of live broadcast is presented to users in the form of platform live broadcast, etc.

[0085] Based on the user personalized data obtained from the offline personalized data processing process, the intelligent solution system uses big model technology to generate solutions with personalized capabilities in real time, such as the form of "audio 1+video 2+conversation 3".

[0086] like Figure 5 As shown, in a more preferred embodiment, step S5, inputting the personalized prompt words corresponding to the user into the large model to generate corresponding personalized reply content, including:

[0087] Step S91, inputting the personalized prompt words corresponding to the user into the macro model to generate corresponding personalized reply sub-content;

[0088] Step S92: Based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language, the personalized reply sub-contents are combined to obtain the corresponding personalized reply content.

[0089] Here, this solution innovatively improves the content satisfaction method in the traditional service form, and upgrades the user unified content method in the traditional service form to one based on user personalization. It can disassemble the learning content, reorganize the information structure in the smallest granularity, and recommend the most suitable content to different users. Since the improvement allows different users to see the content that suits them best, it produces higher learning efficiency for users.

[0090] AI and LLM technology can be used to break down content into the smallest information bodies. Each smallest information body only includes atomic information. For example, the content "How to be a good mother" can be broken down into related sub-content such as "Necessary conditions for a good mother" and "How to better understand children". The sub-content can be combined based on the user's personalized information to obtain corresponding personalized reply content, thereby meeting the user's learning needs in the most efficient way.

[0091] The above-mentioned contents in the form of text, video and audio are organized in a manner of minimum granularity information, which can break up the original coarse-grained information.

[0092] like Figure 3 As shown, in a preferred embodiment, after step S5 generates the corresponding personalized reply content, it also includes:

[0093] Step S71, if the personalized reply content is in a dialogue mode, the user role customization system defines the user role;

[0094] Step S72, the AI ​​role customization system is used to define the AI ​​role that communicates with the user;

[0095] Step S73, the AI ​​dialogue system constructs a virtual AI character based on the defined user character and AI character to practice dialogue with the user;

[0096] In step S74, the AI ​​analysis system analyzes the content of each dialogue exercise, points out what the user did well and what needs improvement, and summarizes and provides feedback to the user in a timely manner.

[0097] Here, the dialogue results obtained by the intelligent solution system can trigger the role-playing practice system. The role-playing practice system includes the user role customization system, the AI ​​role customization system, the AI ​​dialogue system, and the AI ​​analysis system.

[0098] In traditional service forms, traditional educational methods such as exercises are used to respond to users' practice demands. This method is too passive and not conducive to users' active thinking. Based on AI LLM technology, the present invention can train a variety of role-playing models. Users can also customize their own role-playing models, assist users in practice through dialogue, analyze the dialogue content after each practice, summarize and feedback to users in a timely manner, and ensure the integration of "learning + practice" for users.

[0099] Specifically, the user role customization system is used to define user roles, that is, a detailed description of the user role during the conversation with AI, such as "You are a 30-year-old mother who is currently experiencing problems with your child's admission to kindergarten." A new user role can be defined for each conversation.

[0100] The AI ​​character customization system is used to define the AI ​​character that communicates with the user. That is, during the conversation with the AI, a detailed description of the AI ​​character is provided. For example, "You are a three-year-old child who is currently experiencing problems adapting to kindergarten." A brand new AI character can be defined for each conversation.

[0101] The AI ​​dialogue system constructs a virtual AI character based on the defined user role and AI role to conduct dialogues with the user in the form of text, voice, etc. The AI ​​dialogue system is implemented by a general large model and a large model obtained after training based on the general large model. The general large model uses the industry open source model. The large model obtained after training based on the general large model is fine-tuned and reinforced with learning training based on our data. We select general large models of different sizes such as 7B parameters, 14B parameters, and 70B parameters for training. After training and testing, the dialogue effect is guaranteed to be accurate.

[0102] After the conversation is over, the AI ​​analysis system analyzes the conversation content between the user and the AI, points out what the user did well and what needs improvement, and presents it to the user in the form of an analysis report.

[0103] The AI ​​analysis system can be implemented by a general large model and a large model obtained after training based on the general large model. The general large model uses the industry open source model. The large model obtained after training based on the general large model is fine-tuned and reinforced with learning training based on our data. We select general large models of different sizes such as 7B parameters, 14B parameters, and 70B parameters for training. After training and testing, the analysis effect is accurate.

[0104] This patent solution innovatively improves the user practice method in the traditional service form, from the traditional exercise question method to an AI-based role-playing dialogue method. Since the improved dialogue form enables users to think more actively, it produces better practice effects for users.

[0105] like Figure 4 As shown, in a more preferred embodiment, step S5, inputs the personalized prompt word corresponding to the user into the large model to generate the corresponding personalized reply content, and then further includes:

[0106] Step S81, the AI ​​test generation system generates corresponding personalized test questions based on the corresponding personalized reply content;

[0107] Step S82: The AI ​​test system presents personalized test questions and obtains the user's answers within a limited time.

[0108] Step S83, the AI ​​test analysis system generates a test analysis report based on the answer content;

[0109] Step S84: The AI ​​test feedback system adjusts the personalized reply content and / or the display method of the personalized reply content based on the test analysis report.

[0110] Here, the AI ​​test system includes modules such as AI test generation system, AI test system, AI test analysis system, and AI test feedback system.

[0111] The AI ​​test generation system can utilize user personalized data and generate personalized test questionnaires based on user daily behavior. The personalized test questionnaire includes multiple test questions, and the test questionnaire and test questions seen by each user are different.

[0112] Using the user's personalized data as input, the big model is informed of prompt words to generate personalized content, and ultimately generates personalized test questions.

[0113] Specifically, the AI ​​testing system may present personalized test questions and obtain the user's answers within a limited time.

[0114] Use AI technology to analyze the answers to personalized test questions completed by users in the AI ​​testing system and generate a test analysis report.

[0115] Based on the test analysis report generated by the AI ​​test analysis system, dynamically adjust the solution in the intelligent solution system.

[0116] The dynamic solution is generated by the intelligent solution system based on the user's personalized data and the test analysis report generated by the AI ​​test analysis system as input to generate a new solution. For example, the original "video + audio" display solution has been adjusted and updated to a "content + dialogue" display solution.

[0117] According to another aspect of the present invention, there is further provided a computer-readable storage medium having computer-executable instructions stored thereon, wherein when the computer-executable instructions are executed by a processor, the processor is caused to: execute any of the methods described above.

[0118] According to another aspect of the present invention, there is also provided a computer device, comprising:

[0119] Processor; and

[0120] A memory arranged to store computer executable instructions, which when executed cause the processor to: perform any of the methods described above.

[0121] The detailed contents of the device embodiments of the present invention can be found in the corresponding parts of the method embodiments, which will not be described in detail here.

[0122] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

[0123] It should be noted that the present invention can be implemented in software and / or a combination of software and hardware, for example, can be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device. In one embodiment, the software program of the present invention can be executed by a processor to implement the steps or functions described above. Similarly, the software program of the present invention (including related data structures) can be stored in a computer readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some steps or functions of the present invention can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform each step or function.

[0124] In addition, a part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. The program instruction for calling the method of the present invention may be stored in a fixed or removable recording medium, and / or transmitted through a data stream in a broadcast or other signal-bearing medium, and / or stored in a working memory of a computer device that runs according to the program instruction. Here, according to an embodiment of the present invention, a device is included, the device including a memory for storing computer program instructions and a processor for executing program instructions, wherein, when the computer program instruction is executed by the processor, the device is triggered to run the method and / or technical solution based on the aforementioned multiple embodiments of the present invention.

[0125] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. A question-answering method based on user personalized data, characterized in that: include: The data return system obtains the user's personalized data; The data return system processes the personalized data and fills it into the structured pattern and unstructured personalized information defined in the personalized schema; The personalized data processing system, based on the personalized data filled into the structured pattern defined by the personalized schema and the unstructured personalized information, obtains the structured schema data defined by the personalized schema and the unstructured personalized data described in natural language through AI processing; Personalized schema management system, which stores structured schema data corresponding to each user and unstructured personalized data described in natural language; The online personalized content generation module obtains the structured schema data corresponding to the user and the unstructured personalized data described in natural language from the personalized schema management system, and matches the corresponding prompt word template in the prompt warehouse based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language; Based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language, the matched prompt word templates are assembled to obtain the personalized prompt words corresponding to the user, and the personalized prompt words corresponding to the user are input into the big model to generate the corresponding personalized reply content.

2. The question-answering method based on user personalized data according to claim 1, characterized in that: The personalized data includes: data on user conversations, user behaviors and user attributes.

3. The question-answering method based on user personalized data as claimed in claim 2, characterized in that: The data return system processes the personalized data and fills it into the structured pattern and unstructured personalized information defined in the personalized schema, including: Extract key information about user intent and user events from the user conversation data, and fill the extracted key information about user intent and user events into the corresponding fields of intent and events in the defined structured pattern of the personalized schema; at the same time, perform relevant processing of summary and description summary from the user conversation data, and fill the summary and description summary into the unstructured personalized information; Extract the user usage time and user stay time behavior data from the user behavior data, and fill the extracted relevant data into the usage time and stay time and other related fields in the defined structured pattern of the personalized schema; at the same time, use the LLM technology to generate a natural language description of the user behavior data and fill it into the unstructured personalized information; Extract age and gender related fields from user profile data and fill them into the age and gender related fields in the structured pattern defined in the personalized schema; at the same time, use LLM technology to generate a natural language description of the user profile data and fill it into the unstructured personalized information.

4. The question-answering method based on user personalized data according to claim 1, characterized in that: In the prompt warehouse, match the corresponding prompt word template. The prompt warehouse includes multiple prompt word templates, each prompt template represents a prompt description of a specific scenario, corresponding structured schema data of different users and unstructured personalized data described in natural language, matching different prompt word templates.

5. The question-answering method based on user personalized data according to claim 1, characterized in that: After generating the corresponding personalized reply content, it also includes: Based on the user's corresponding structured schema data and unstructured personalized data described in natural language, personalized reply content is displayed in one or more ways: text, pictures, videos, conversations, and live broadcasts.

6. The question-answering method based on user personalized data according to claim 1, characterized in that: The personalized prompt words corresponding to the user are input into the big model to generate the corresponding personalized reply content, including: Input the personalized prompt words corresponding to the user into the large model to generate corresponding personalized reply sub-content; Based on the structured schema data corresponding to the user and the unstructured personalized data described in natural language, the personalized reply sub-contents are combined to obtain the corresponding personalized reply content.

7. The question-answering method based on user personalized data as claimed in claim 5, characterized in that: After generating the corresponding personalized reply content, it also includes: If the personalized reply content is in a conversational manner, the user role customization system defines the user role; The AI ​​character customization system is used to define AI characters that communicate with users; The AI ​​dialogue system constructs a virtual AI character based on the defined user role and AI role to practice dialogue with the user; The AI ​​analysis system analyzes the content of each conversation exercise, points out what the user did well and what needs improvement, and provides timely summaries and feedback to the user.

8. The question-answering method based on user personalized data according to claim 1, characterized in that: The personalized prompt words corresponding to the user are input into the big model to generate the corresponding personalized reply content, and then it also includes: The AI ​​test generation system generates corresponding personalized test questions based on the corresponding personalized response content; The AI ​​test system presents personalized test questions and obtains the user's answers within a limited time. The AI ​​test analysis system generates a test analysis report based on the answer content; The AI ​​test feedback system adjusts the personalized response content and / or the display method of the personalized response content based on the test analysis report.

9. A computer-readable storage medium having computer-executable instructions stored thereon, wherein: When the computer executable instructions are executed by a processor, the processor is caused to: perform the method according to any one of claims 1 to 8.

10. A computer device, wherein: include: processor; as well as A memory arranged to store computer executable instructions which, when executed, cause the processor to: perform a method as claimed in any one of claims 1 to 8.

Images