Customized intelligent question and answer and service method and system fusing large language model

Abstract

The invention provides a customized intelligent question answering and service method and system fusing a language large model, and the method comprises the steps: obtaining the special data information of a user, and constructing the user portrait information based on the special data information; training based on the proprietary data information to obtain a plurality of data pushing models; in response to demand information, determining an intention data pushing model from a plurality of data pushing models based on the demand information and the user portrait information; generating push information based on the demand information and the intention data push model; and pushing the push information based on a preset push mode. Therefore, in the question and answer and service process of the user, the push information meeting the requirements of the user can be generated, and the use experience of the user is improved.

Description

Technical Field

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a customized intelligent question-answering and service method and system integrating a large language model. Background Art

[0002] With the rise of AI big model technology, there will be a boom in research and application of intelligent services based on massive data and big models, generative data analysis, intelligent question and answer, and various scenarios. However, there is still a relatively blank area in terms of intelligent services serving fields, groups, organizations, and individuals. Intelligent services cannot provide users with customized questions and answers and services, which reduces the user experience. Summary of the invention

[0003] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0004] To this end, the first purpose of this application is to propose a customized intelligent question-answering and service method that integrates a large language model.

[0005] The second purpose of this application is to propose a customized intelligent question-answering and service system that integrates a large language model.

[0006] The third objective of the present application is to provide an electronic device.

[0007] A fourth objective of the present application is to provide a computer-readable storage medium.

[0008] To achieve the above-mentioned objectives, the first aspect of the present application proposes a customized intelligent question-answering and service method that integrates a large language model, including: obtaining the user's proprietary data information, and constructing user portrait information based on the proprietary data information; training a plurality of data push models based on the proprietary data information; in response to demand information, determining an intended data push model from a plurality of data push models based on the demand information and the user portrait information; generating push information based on the demand information and the intended data push model; and pushing the push information based on a preset push mode and the user portrait information.

[0009] According to one embodiment of the present disclosure, the obtaining of user proprietary data information includes: obtaining system data and user data as internal data; determining a user domain, and obtaining public data based on the user domain and public resources as external data; determining model information of a preset model in the system, and annotating the internal data and the external data based on the model information to obtain proprietary data information.

[0010] According to one embodiment of the present disclosure, constructing user portrait information based on the proprietary data information includes: obtaining user basic information based on the user data; obtaining user behavior information based on the system data; extracting user feature information from the user basic information and the user behavior information; clustering the user feature information to obtain different user groups; and analyzing the user groups to construct user portrait information for the user groups.

[0011] According to one embodiment of the present disclosure, the multiple data push models are obtained based on the training of the proprietary data information, including: obtaining the data type of the proprietary data information, and dividing the proprietary data information into data groups according to the data type; determining the training input data and the training output data in each of the data groups; inputting the training input data and the training output data into a preset model for model training, and obtaining the data push model corresponding to the data group.

[0012] According to one embodiment of the present disclosure, determining an intended data push model from a plurality of data push models based on the demand information and the user portrait information includes: extracting first feature information based on the demand information, and being able to obtain a data domain based on the first feature information; extracting second feature information based on the user portrait information, and being able to obtain user intention information based on the second feature information; and determining an intended data push model based on the data domain and the user intention information.

[0013] According to one embodiment of the present disclosure, the push information is generated based on the demand information and the intention data push model, including: determining model input data based on the demand information; inputting the model input data into the intention data push model to obtain alternative push information; sending the alternative push information to an expert review end to obtain a review result; if the review result is passed, using the alternative push information as the push information.

[0014] According to one embodiment of the present disclosure, the pushing of the push information based on a preset push mode includes: obtaining a preset push mode, the push mode including a preset mode and a customized mode; in response to the demand information, obtaining an information receiving end and an information display method; and pushing the push information based on the push mode, the information receiving end and the information display method.

[0015] To achieve the above-mentioned purpose, the second aspect embodiment of the present application proposes a customized intelligent question and answer and service system that integrates a large language model, including a data acquisition module, a model training module, a model selection module and an information push module; the data acquisition module is used to acquire the user's proprietary data information and construct user portrait information based on the proprietary data information; the model training module is used to train a variety of data push models based on the proprietary data information; the model selection module is used to respond to demand information and determine an intended data push model from a variety of data push models based on the demand information and the user portrait information; the information push module is used to generate push information based on the demand information and the intended data push model, and push the push information based on a preset push mode.

[0016] To achieve the above-mentioned purpose, the third aspect embodiment of the present application proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method as described above.

[0017] To achieve the above-mentioned purpose, the fourth aspect embodiment of the present application proposes a computer-readable storage medium, in which computer-readable storage medium is stored computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method described above.

[0018] The present application provides a customized intelligent question-answering and service method and system that integrates a large language model. By acquiring the user's proprietary data information, a customized portrait is created for the user, and user portrait information is constructed. Thus, an intention data push model is provided for the user based on the user's demand information and user portrait information, thereby generating push information that meets the user's needs and improving the user's usage experience.

[0019] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0021] Figure 1 A flowchart of a customized intelligent question-answering and service method integrating a large language model provided in an embodiment of the present application;

[0022] Figure 2 This is a flow chart of steps S201-S203 of a customized intelligent question-answering and service method integrating a large language model according to an embodiment of the present application;

[0023] Figure 3 This is a flow chart of steps S301-S305 of a customized intelligent question-answering and service method integrating a large language model according to an embodiment of the present application;

[0024] Figure 4 This is a flow chart of steps S401-S403 of a customized intelligent question-answering and service method integrating a large language model according to an embodiment of the present application;

[0025] Figure 5 This is a flow chart of steps S501-S503 of a customized intelligent question-answering and service method integrating a large language model according to an embodiment of the present application;

[0026] Figure 6 This is a flow chart of steps S601-S604 of a customized intelligent question-answering and service method integrating a large language model according to an embodiment of the present application;

[0027] Figure 7 This is a flowchart of steps S701-S703 of a customized intelligent question-answering and service method integrating a large language model according to an embodiment of the present application;

[0028] Figure 8 A module diagram of a customized intelligent question-answering and service system that integrates a large language model provided in an embodiment of the present application.

[0029] Description of reference numerals:

[0030] 1. Data acquisition module; 2. Model training module; 3. Model selection module; 4. Information push module. DETAILED DESCRIPTION

[0031] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0032] The acquisition, storage, use, and processing of data in the technical solution disclosed in this embodiment comply with relevant provisions of national laws and regulations.

[0033] The following describes a customized intelligent question-answering and service method and system that integrates a large language model in an embodiment of the present application with reference to the accompanying drawings.

[0034] Figure 1 The following is a flow chart of a customized intelligent question-answering and service method integrating a large language model provided in an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:

[0035] Step S101, obtaining user's proprietary data information, and constructing user portrait information based on the proprietary data information.

[0036] In one implementation of an embodiment of the present application, the customized intelligent question and answer and service method of the fusion language big model can be applied to the scenario where the user asks questions on the terminal. The platform adopted can be the customized intelligent question and answer and service system of the fusion language big model provided in the embodiment of the present application. If it is deployed at the group company level, it can provide users with intelligent question and answer, intelligent customer service, intelligent retrieval, opportunity identification, risk warning, cooperation recommendation, information push, statistical summary, document preparation and other intelligent services on the PC terminal, mobile terminal and group large screen of the group's internal and external networks.

[0037] In this embodiment, the user group is a group, organization or individual, and the user's proprietary data information refers to the proprietary information belonging to the user generated in the process of using the system. In one implementation of this embodiment, the proprietary data information may include multiple types, without any limitation here, such as user identity information, registration information, login information, browsing information and search information, socially disclosed data, Internet of Things data, industry knowledge and industry reports, etc.

[0038] After obtaining the user's proprietary data information, the user portrait information can be constructed based on the proprietary data information. The user portrait information can reflect the user's interests, preferences, areas of expertise, usage habits, etc.

[0039] Optionally, refer to Figure 2 , the method for obtaining the user's proprietary data may also include:

[0040] S201. Obtain system data and user data as internal data.

[0041] Among them, user data refers to the user's identity information, etc., and system data refers to registration information, login information, browsing information, and search information, etc. When using the system, users need to register and log in. During this process, the user's identity information needs to be recorded. In the subsequent use of the system, the browsing behavior, search behavior, and question-and-answer behavior generated will be recorded by the system, thus forming system data.

[0042] Due to the privacy of user data and system data, it is necessary to use encryption transmission technology to encrypt and transmit internal data. There are many types of encryption transmission technology, and this embodiment does not limit the encryption transmission technology, for example, blockchain technology and privacy computing technology. Among them, blockchain is a distributed database that makes data traceable and auditable in a decentralized manner without being tampered with. In blockchain technology, data is open and transparent, but the identity information of traders is anonymous, which makes blockchain technology have high privacy protection. By using public and private keys for encryption and decryption, transactions can remain anonymous and secure. At the same time, blockchain technology can be used for data retrieval, authorization, authentication and other capabilities, and can provide tamper-proof data flow logs when tracing back afterwards. As for privacy computing, it can achieve the effect of data being available but invisible, and data being calculable but unrecognizable. For example, multi-party joint computing, multi-party joint modeling, and privacy query are performed without the original data being stored, so as to achieve the effect of "data is not moving, computing is moving, data is available but invisible".

[0043] S202: Determine the user domain, and obtain public data based on the user domain and public resources as external data.

[0044] Among them, external data can be obtained through public resources, and the acquisition is targeted according to the user's field. For example, if the user belongs to the energy industry, relevant data can be obtained within the energy industry. The data can be obtained through web crawlers, or directly using the data already available from the data model manufacturer, or in the form of industry, organization, report, etc.

[0045] Of course, different external data can be obtained according to different user fields for application in the data push model.

[0046] S203: Determine model information of a preset model in the system, and annotate internal data and external data based on the model information to obtain proprietary data information.

[0047] In this embodiment, a variety of preset models are stored in the system. To facilitate the training and use of the preset models, data annotation processing can be performed on the data required by the model for the model information of the preset model, that is, adding labels or annotations to the data, thereby obtaining dedicated data information that can be directly used by the preset model. After obtaining the internal data and external data, data annotation work can be performed manually to improve the availability, richness and accuracy of the internal data and external data.

[0048] Reference Figure 3 In the process of obtaining user portrait information according to the dedicated data information, the following steps are also included:

[0049] S301. Obtain basic user information based on user data.

[0050] The user data includes basic user information, such as identity information, etc., which is not limited in this embodiment.

[0051] S302: Obtain user behavior information based on system data.

[0052] User behavior information includes the above-mentioned registration information, login information, browsing information, search information, and question and answer information.

[0053] S303: extracting user feature information from user basic information and user behavior information.

[0054] Clean and deduplicate the data in user basic information and user behavior information, remove invalid and erroneous data, and use data mining, machine learning and other technologies to extract user characteristic information, such as occupation, age, gender, interests and hobbies, so as to facilitate user profiling based on user characteristic information.

[0055] S304: Perform cluster analysis on user feature information to obtain different user groups and group labels.

[0056] Integrate the extracted user feature information, and use cluster analysis, association rules and other methods to divide users into different groups or categories, and assign group labels to each group or category. These group labels can represent the characteristics and preferences of users. For example, for young women, labels such as fashion, beauty, and food can be assigned.

[0057] S305: Build user portrait information of users based on user groups and group labels.

[0058] By integrating the user group to which each user belongs and the group label corresponding to the user group, the user label of the user can be obtained, so that the user can be profiled according to the user label to obtain user portrait information. The user portrait information may include the user's occupation, interests, hobbies, behavioral habits, etc.

[0059] Step S102, multiple data push models are obtained through training based on proprietary data information.

[0060] In this embodiment, the preset model is trained according to the user's proprietary data information, so as to provide the user with a data push model that meets the user's needs.

[0061] In one implementation of this embodiment, training data that facilitates the training of a preset model is constructed based on the user's proprietary data information, so that the preset model can be trained.

[0062] Optionally, in this embodiment, there is no restriction on the type of preset model, and any language model that can be applied to the customized intelligent question-answering and service system can be used as a preset model, for example, BERT, GPT, convolutional neural network (CNN), recurrent neural network (RNN) and other language models. By training the language model, a data push model that meets user needs can be obtained.

[0063] Optionally, the data push model in this embodiment can be multiple, and targeted analysis and calculation are performed on different fields and industries to improve the accuracy of the output results. It is understandable that data push models with different specialties are trained based on user-specific data information with different characteristics, so that users can flexibly choose them, so that the output results are closer to user needs.

[0064] Reference Figure 4 , training multiple data push models based on proprietary data information may include the following steps:

[0065] S401. Obtain data types of proprietary data information, and divide the proprietary data information into data groups according to the data types.

[0066] The proprietary data information is divided into data types. In this embodiment, the data types may correspond to the above group labels. For different data types, adaptive clustering is performed to obtain different types of data groups. This application does not impose any restrictions on the data types.

[0067] S402, determining the training input data and the training output data in each data group;

[0068] After obtaining the data group, the training input data and the training output data can be divided in the data group of the same data type. Optionally, the test data can also be divided from the data group, and the preset model can be trained by the training input data and the training output data, and the accuracy of the trained model can be tested by the test data.

[0069] S403: Input the training input data and the training output data into a preset model for model training to obtain a data push model corresponding to the data group.

[0070] Optionally, in this embodiment, after the preset model is trained according to the training input data and the training output data, the trained model is tested with the test data until the model meets the accuracy requirements, thereby obtaining the data push model corresponding to the data group.

[0071] It is understandable that, since different data push models are suitable for different data types, when in use, a data push model suitable for the user's intention can be obtained based on the user's demand information analysis, thereby providing the user with customized information push.

[0072] Step S103, in response to the demand information, determining an intended data push model from a plurality of data push models based on the demand information and the user portrait information.

[0073] Users send out demand information in the customized intelligent question-and-answer and service system, such as searching for information, asking questions, etc. The system determines the data push model suitable for the user based on the user's demand information and user portrait information, so as to provide users with output results that meet their needs.

[0074] Optionally, refer to Figure 5 , determining the intended data push model from multiple data push models based on the demand information and the user's Huaxi information, and further comprising the following steps:

[0075] S501. Extract first characteristic information based on demand information, and acquire a data domain based on the first characteristic information.

[0076] In this embodiment, the first characteristic information is extracted based on the user's demand information. The first characteristic information may be text or pictures with iconic content in the demand information. For example, if the demand information input by the user in the system is "how to prevent colds", "colds" and "prevention" are extracted as the first characteristic information. Based on the first characteristic information, it can be determined that the data field is the medical field.

[0077] Optionally, this embodiment does not impose any restriction on the data field, so when determining the data field according to the first characteristic information, multiple data fields related to the demand information can be obtained at the same time.

[0078] S502: extract second feature information based on the user portrait information, and obtain user intention information based on the second feature information.

[0079] In this embodiment, the second characteristic information is extracted based on the user portrait, and the second characteristic information may be the user tag mentioned above. According to the second characteristic information, the user's intention information that the user is more interested in can be obtained. For example, if the user tag in the user portrait information indicates that the user is an elderly person, the user tag of "elderly person" is used as the second characteristic information, and the user can be determined to be an elderly person according to the second characteristic information.

[0080] Optionally, this embodiment also does not impose any restrictions on user intention information. Therefore, when determining user intention information based on the second characteristic information, multiple user intention information related to user portrait information can be obtained at the same time, for example, user intention information includes "female", "elderly", etc.

[0081] S503: Determine an intention data push model based on the data domain and user intention information.

[0082] After determining the data domain and user intention information, the intended data push model can be determined from all data push models based on the data domain and user intention information. For example, when the data domain is determined to be the medical field and the user intention information is "elderly", prediction models such as medical and elderly diseases can be selected as intended data push models, so that customized results can be output based on the user's demand information.

[0083] Step S104: Generate push information based on the demand information and the intention data push model.

[0084] In this embodiment, after determining the intention data push model, the demand information is adjusted to the format supported by the intention data push model, and the intention data push model is input. The intention data push model can generate output results as push information for users to view and understand.

[0085] Reference Figure 6 , generating push information based on demand information and intention data push model, can also include the following steps:

[0086] S601. Determine model input data based on demand information.

[0087] Before the demand information is input into the intention data push model, the demand information needs to be preprocessed to form model input data that is easy for the intention data push model to recognize. In this embodiment, feature information, namely first feature information, is extracted according to the demand information, and the model input data can be determined according to the first feature information.

[0088] S602: Input the model input data into the intention data push model to obtain the candidate push information.

[0089] The model input data is input into the intention data push model, and the intention data push model generates output results after analysis and calculation. After the output results are sorted, the candidate push information can be obtained, for example, the candidate push information is medicines, foods and exercises suitable for the elderly. In this embodiment, the candidate push information can be the output results to be reviewed.

[0090] S603: Send the candidate push information to the expert review terminal to obtain the review result.

[0091] In order to improve the accuracy of the pushed information, the alternative push information is transmitted to the expert review end, which is used to review the alternative push information based on the demand information and user portrait information, and output the review results.

[0092] S604: If the review result is passed, the alternative push information is used as the push information.

[0093] When the review result is passed, it is determined that the candidate push information can be pushed to the user, and the candidate push information is regarded as the push information that meets the requirements.

[0094] Step S106: Push the push information based on the preset push mode.

[0095] In this embodiment, different push modes are stored in the system. The push mode is determined according to the user's usage scenario or user needs. After the push mode is determined, information can be pushed to the user according to the push mode.

[0096] Optionally, refer to Figure 7 , push the push information based on the preset push mode, including the following steps:

[0097] S701, obtaining a preset push mode, where the push mode includes a preset mode and a customized mode;

[0098] S702, in response to the demand information, obtaining an information receiving terminal and an information display method;

[0099] S703: Push information based on the push mode, information receiving terminal and information display method.

[0100] In this embodiment, the push mode includes a preset mode and a customized mode, wherein the preset mode is a preset service such as intelligent question and answer, intelligent customer service, intelligent search, opportunity identification, risk warning, cooperation recommendation, information push, statistical summary, and document compilation provided by the system based on the needs of the majority of people and the system's special needs. The customized mode is a special service provided based on the preset needs or demand information input by the user, user portrait information, etc.

[0101] After determining the push mode, the information receiving terminal and the information display method can be determined according to the demand information. In this embodiment, there is no restriction on the information receiving terminal and the information display method. For example, the information receiving terminal includes a computer terminal, a display large screen terminal, a mobile phone terminal, a watch terminal, a speaker terminal, a headset terminal and other smart device terminals, and the information display method includes text, voice, pictures, etc.

[0102] So far, the push mode, information receiving terminal and information display method have been determined. In order to achieve the timeliness of the service, the push information can be automatically pushed to the user's information receiving terminal.

[0103] Optionally, the system's push results can also be combined with virtual people, such as providing relevant result output and broadcast publicity through the group's proprietary digital virtual people and intelligent customer service virtual people. For example, the group's digital virtual people can broadcast preset business statistics, opportunity analysis, risk warning, cooperation recommendations, internal and external news, scientific and technological progress, popular science knowledge, corporate publicity, etc. on the group's large screen in real time to enhance the vividness of information interaction.

[0104] In order to implement the above embodiments, the present application also proposes a customized intelligent question-answering and service system that integrates a large language model.

[0105] Figure 8 A module diagram of a customized intelligent question-answering and service system integrating a large language model provided in an embodiment of the present application.

[0106] like Figure 8 As shown, the system includes a data acquisition module 1, a model training module 2, a model selection module 3 and an information push module 4.

[0107] The data acquisition module 1 is used to acquire the user's proprietary data information and construct user portrait information based on the proprietary data information;

[0108] The model training module 2 is used to train a variety of data push models based on proprietary data information;

[0109] The model selection module 3 is used to determine the intended data push model from multiple data push models based on the demand information and user portrait information in response to the demand information;

[0110] The information push module 4 is used to generate push information based on the demand information and the intention data push model, and push the push information based on a preset push mode.

[0111] It should be noted that the above explanation of the embodiment of the customized intelligent question and answer and service method for the fusion language big model is also applicable to the customized intelligent question and answer and service system for the fusion language big model of this embodiment, and will not be repeated here.

[0112] In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the customized intelligent question-answering and service method of the fused language large model provided in the above embodiments.

[0113] In order to implement the above embodiments, the embodiments of the present disclosure also propose a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to implement a customized intelligent question-answering and service method of the fused language big model as in the first aspect of the embodiment of the present disclosure.

[0114] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this application are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0115] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0116] The present application is expected to provide an implementation scheme for users to selectively block the use or access of personal information data. That is, the present disclosure is expected to provide hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, the risk can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0117] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0118] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0119] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0120] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0121] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0122] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0123] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0124] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A customized intelligent question-answering and service method integrating a large language model, characterized in that: include: Acquire the user's proprietary data information, and construct user profile information based on the proprietary data information; Based on the proprietary data information, a plurality of data push models are trained; In response to the demand information, determining an intended data push model from a plurality of data push models based on the demand information and the user portrait information; Generate push information based on the demand information and the intention data push model; The push information is pushed based on a preset push mode.

2. The method according to claim 1, characterized in that The obtaining of user's proprietary data information includes: Obtain system data and user data as internal data; Determine a user domain, and obtain public data based on the user domain and public resources as external data; Model information of a preset model in the system is determined, and data annotation is performed on the internal data and the external data based on the model information to obtain proprietary data information.

3. The method according to claim 2, characterized in that The constructing user portrait information based on the proprietary data information includes: Acquire basic user information based on the user data; Acquire user behavior information based on the system data; Extracting user feature information from the user basic information and the user behavior information; Clustering the user characteristic information to obtain different user groups; The user group is analyzed to construct user portrait information of the user group.

4. The method according to claim 2, characterized in that: The multiple data push models are obtained by training based on the proprietary data information, including: Acquiring a data type of the proprietary data information, and dividing the proprietary data information into data groups according to the data type; determining training input data and training output data in each of the data groups; The training input data and the training output data are input into a preset model for model training to obtain a data push model corresponding to the data group.

5. The method according to claim 1, characterized in that The determining of the intended data push model from the multiple data push models based on the demand information and the user portrait information includes: Extracting first characteristic information based on the demand information, and acquiring a data domain based on the first characteristic information; Extracting second feature information based on the user portrait information, and acquiring user intention information based on the second feature information; An intention data push model is determined based on the data domain and the user intention information.

6. The method according to claim 1, characterized in that The generating the push information based on the demand information and the intention data push model includes: Determining model input data based on the demand information; Inputting the model input data into the intention data push model to obtain candidate push information; Send the candidate push information to the expert review terminal to obtain the review result; If the review result is passed, the candidate push information is used as the push information.

7. The method according to claim 1, characterized in that The pushing the push information based on a preset push mode includes: Obtaining a preset push mode, wherein the push mode includes a preset mode and a customized mode; In response to the demand information, obtaining an information receiving terminal and an information display method; The push information is pushed based on the push mode, information receiving terminal and information display method.

8. A customized intelligent question-answering and service system integrating a large language model, characterized in that: It includes a data acquisition module (1), a model training module (2), a model selection module (3) and an information push module (4); The data acquisition module (1) is used to acquire the user's proprietary data information and construct user portrait information based on the proprietary data information; The model training module (2) is used to train a plurality of data push models based on the proprietary data information; The model selection module (3) is used to determine the intended data push model from the multiple data push models based on the demand information and the user portrait information in response to the demand information; The information push module (4) is used to generate push information based on the demand information and the intention data push model, and to push the push information based on a preset push mode.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images