Insurance recommendation method and device, equipment and medium

Abstract

The invention belongs to the field of big data, and relates to an insurance recommendation method and device, equipment and a medium, and the method comprises the steps: obtaining online behavior data and offline behavior data of a user; integrating the online behavior data and the offline behavior data through a preset multi-source data fusion engine, and constructing a user portrait of the user; based on the user portrait, generating a demand prediction result of the user through a preset privacy protection calculation framework; based on the demand prediction result, displaying dynamic demonstration content of the virtual insurance scene to the user through a preset scene interaction module to obtain interaction input data of the user; acquiring real-time behavior data of the user, and adjusting an insurance scheme of the user based on the user portrait, the interactive input data and the real-time behavior data to obtain a to-be-recommended insurance scheme; and based on the to-be-recommended insurance scheme, performing insurance recommendation to the user. The method can be applied to the business fields of financial insurance and the like, and insurance recommendation accuracy can be improved.

Description

Technical Field

[0001] This application relates to the field of big data technology and is applied to online processing business scenarios such as financial insurance, and in particular to an insurance recommendation method, device, equipment and medium. Background Technology

[0002] Traditional insurance recommendation methods heavily rely on human experience and simple user tags. While human recommendations leverage the expertise and experience of professionals, their inherent limitations in practice stem from the inherent constraints of individual knowledge and experience. Faced with the increasingly complex and ever-changing user needs in the insurance market, human intervention struggles to provide comprehensive coverage and accurate insights. Furthermore, simple user tags often only offer a rough categorization of users from a single dimension or a few aspects, failing to capture the nuanced characteristics and diverse potential needs of users. This results in difficulties in accurately matching users with insurance products, significantly reducing the efficiency of matching user needs and leaving many potential insurance requirements unmet.

[0003] Even more serious is the significant gap in online-offline collaboration. Offline insurance sales scenarios function like isolated information islands, with severely hampered data flow between them and online platforms. Detailed user information acquired offline cannot be synchronized to the online system in a timely and smooth manner, resulting in a waste of data resources. Simultaneously, the abundant online data resources are difficult to effectively feed back into offline service processes, failing to provide strong support for offline sales. This data fragmentation prevents insurance services from forming a complete data loop, lacking effective linkage and collaboration between various stages, greatly limiting the overall efficiency of insurance services and making it difficult to provide users with a seamless, efficient, and convenient insurance experience.

[0004] In conclusion, traditional insurance recommendations are inefficient in matching user needs and suffer from a severe lack of online-offline coordination, making them unsuitable for the development needs of the insurance industry. Summary of the Invention

[0005] The purpose of this application is to provide an insurance recommendation method, apparatus, computer equipment, and storage medium to solve the problems of low efficiency in matching user needs and serious lack of online and offline collaboration in traditional insurance recommendations.

[0006] Firstly, an insurance recommendation method is provided, which adopts the following technical solution:

[0007] The system acquires the user's online and offline behavior data; integrates the online and offline behavior data using a pre-defined multi-source data fusion engine to construct a user profile; based on the user profile, generates a demand prediction result for the user using a pre-defined privacy-preserving computing framework; based on the demand prediction result, displays a dynamic demonstration of a virtual insurance scenario to the user using a pre-defined scene interaction module to obtain the user's interactive input data; acquires the user's real-time behavior data, and adjusts the user's insurance plan based on the user profile, the interactive input data, and the real-time behavior data to obtain a recommended insurance plan; and recommends insurance to the user based on the recommended insurance plan.

[0008] Secondly, an insurance recommendation device is provided, which adopts the following technical solution:

[0009] The acquisition module is used to acquire users' online and offline behavior data;

[0010] The integration module is used to integrate the online behavior data and the offline behavior data through a preset multi-source data fusion engine to construct the user profile of the user;

[0011] The generation module is used to generate the user's demand prediction results based on the user profile and through a preset privacy-preserving computing framework.

[0012] The display module is used to present dynamic demonstration content of a virtual insurance scenario to the user through a preset scene interaction module based on the demand prediction results, so as to obtain the user's interactive input data.

[0013] The adjustment module is used to acquire the user's real-time behavior data, and adjust the user's insurance plan based on the user profile, the interaction input data, and the real-time behavior data to obtain the insurance plan to be recommended.

[0014] The recommendation module is used to recommend insurance to the user based on the insurance plan to be recommended.

[0015] Thirdly, a computer device is provided, which adopts the following technical solution:

[0016] The system acquires the user's online and offline behavior data; integrates the online and offline behavior data using a pre-defined multi-source data fusion engine to construct a user profile; based on the user profile, generates a demand prediction result for the user using a pre-defined privacy-preserving computing framework; based on the demand prediction result, displays a dynamic demonstration of a virtual insurance scenario to the user using a pre-defined scene interaction module to obtain the user's interactive input data; acquires the user's real-time behavior data, and adjusts the user's insurance plan based on the user profile, the interactive input data, and the real-time behavior data to obtain a recommended insurance plan; and recommends insurance to the user based on the recommended insurance plan.

[0017] Fourthly, a computer-readable storage medium is provided, which adopts the following technical solution:

[0018] The system acquires the user's online and offline behavior data; integrates the online and offline behavior data using a pre-defined multi-source data fusion engine to construct a user profile; based on the user profile, generates a demand prediction result for the user using a pre-defined privacy-preserving computing framework; based on the demand prediction result, displays a dynamic demonstration of a virtual insurance scenario to the user using a pre-defined scene interaction module to obtain the user's interactive input data; acquires the user's real-time behavior data, and adjusts the user's insurance plan based on the user profile, the interactive input data, and the real-time behavior data to obtain a recommended insurance plan; and recommends insurance to the user based on the recommended insurance plan.

[0019] Compared with existing technologies, the embodiments of this application have the following main advantages: Firstly, by integrating online and offline behavioral data to construct user profiles, the limitations of a single data source are overcome, enabling accurate characterization of complex user features and potential needs. This solves the problems of incomplete coverage and coarse tagging in manual recommendations, leading to inefficient matching. Secondly, the demand prediction mechanism based on a privacy-preserving computing framework achieves an intelligent upgrade in demand insight while ensuring user data security, avoiding excessive reliance on personal experience in traditional methods. The scenario interaction module dynamically demonstrates virtual insurance scenarios, not only enhancing user engagement but also capturing interactive input data in real time and dynamically adjusting insurance plans based on real-time behavioral data. The final output recommendation plan overcomes the lack of coordination caused by the fragmentation of online and offline data and, through multiple rounds of data-driven plan iteration, significantly improves the accuracy of demand matching and insurance service efficiency, thereby enhancing the accuracy of insurance recommendations. Attached Figure Description

[0020] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0022] Figure 2 A flowchart of an embodiment of the insurance recommendation method according to this application;

[0023] Figure 3 This is a schematic diagram of one embodiment of the insurance recommendation device according to this application;

[0024] Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0026] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0027] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0028] like Figure 1As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Terminal device 101 may be a laptop 1011, tablet 1012, or mobile phone 1013. Network 102 is used as a medium to provide a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables.

[0029] Users can use terminal device 101 to interact with server 103 via network 102 to receive or send messages, etc. Various communication client applications can be installed on terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0030] Terminal device 101 can be various electronic devices with a display screen and support web browsing. In addition to laptop computer 1011, tablet computer 1012 or mobile phone 1013, terminal device 101 can also be e-book reader, MP3 player (Moving Picture Experts Group Audio Layer III), MP4 player (Moving Picture Experts Group Audio Layer IV), laptop computer and desktop computer, etc.

[0031] Server 103 can be a server that provides various services, such as a backend server that provides support for the pages displayed on terminal device 101.

[0032] It should be noted that the insurance recommendation method provided in this application embodiment is generally executed by a server / terminal device, and correspondingly, the insurance recommendation device is generally set in the server / terminal device.

[0033] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0034] Continue to refer to Figure 2 A flowchart of an embodiment of the insurance recommendation method according to this application is shown. The insurance recommendation method includes the following steps:

[0035] Step S201: Obtain the user's online behavior data and offline behavior data.

[0036] Here, "user" refers to a specific individual for whom insurance products need to be matched during the insurance service process. This user information originates from registration information in the insurance business system, offline service contact records, or user identifiers from third-party partner platforms. For example, users browsing insurance products through an app or customers consulting at offline branches.

[0037] Online behavioral data refers to the digital behavioral records generated by users on internet insurance service platforms. These records are derived from interactive data such as operation paths and dwell time collected by the application's preset tracking technology, as well as structured information such as web browsing logs, search keywords, and insurance plan collection records.

[0038] Offline behavioral data refers to non-digital behavioral records generated by users in physical insurance service scenarios. These records originate from behavioral trajectory data collected by IoT devices (such as in-vehicle OBD and smart bracelets) and gateway devices, as well as traditional service contact information such as offline interview records and paper questionnaires.

[0039] Step S202: The online behavior data and the offline behavior data are integrated through a preset multi-source data fusion engine to construct the user profile of the user.

[0040] Among them, the multi-source data fusion engine refers to the software module used to integrate online and offline behavioral data. It represents the technological carrier for cross-channel data integration and is used to eliminate data silos between online and offline channels.

[0041] Integration refers to the correlation and unified processing of online and offline behavioral data through a multi-source data fusion engine. It represents the technical process of data fusion and is used to build the foundation for structured user profile data.

[0042] User profiles refer to digital user characteristic models built based on integrated multi-source data, including basic attributes (such as age and occupation), behavioral characteristics (such as online browsing preferences and offline consultation history), and demand tags (such as insurance type and coverage preference), among other multi-dimensional information. They represent a complete digital picture of the user and are used to support accurate demand prediction.

[0043] Step S203: Based on the user profile, generate the user's demand prediction result through a preset privacy-preserving computing framework.

[0044] The privacy-preserving computation framework refers to the technical architecture that ensures user data security during demand forecasting. It can employ techniques such as federated learning, differential privacy, and secure multi-party computation to achieve data usability without visibility. It characterizes the technical boundaries of secure data processing to comply with privacy regulations.

[0045] Among them, the demand forecast result refers to the insurance demand forecast conclusion output by the machine learning model based on user profiles, which may include structured information such as demand type, coverage amount, and product preference.

[0046] Step S204: Based on the demand prediction results, a dynamic demonstration of a virtual insurance scenario is displayed to the user through a preset scenario interaction module to obtain the user's interactive input data.

[0047] The scene interaction module refers to the software system used to enable users to interact with virtual insurance scenes, integrating technical components such as 3D modeling, gesture recognition technology, and voice interaction.

[0048] Virtual insurance scenarios refer to simulated insurance application scenarios built through digital technology, including themes such as family protection planning, vehicle risk demonstration, and health management simulation. For example, a "virtual driving scenario" can be built to demonstrate the changes in the probability of car insurance claims corresponding to different driving habits.

[0049] Among them, dynamic demonstration content refers to the visual content with real-time interactive capabilities generated by the scene interaction module, which can be realized using technologies such as 3D modeling, animation demonstration, and data visualization.

[0050] Interactive input data refers to the operational data generated by users through gestures, voice, and other means while watching dynamic demonstrations of virtual insurance scenarios. For example, if a user asks by voice, "Does this plan include reimbursement for medical treatment abroad?", the record of this question is interactive input data.

[0051] Step S205: Obtain the user's real-time behavior data; based on the user profile, the interaction input data, and the real-time behavior data, adjust the user's insurance plan to obtain the insurance plan to be recommended.

[0052] Real-time behavioral data refers to user behavior information collected in real time through IoT devices, which may include health monitoring data (such as heart rate and sleep data from smart bracelets) and driving behavior data (such as the frequency of emergency braking by in-vehicle OBD devices). It characterizes the risk profile of the user's current state and is used to dynamically adjust insurance plans.

[0053] Among them, insurance plans refer to combinations of insurance products generated based on user profiles, interactive input data, and real-time behavioral data, to meet users' dynamic needs. For example, a combination of "critical illness insurance + term life insurance + medical insurance" might be generated for users who frequently stay up late, with the coverage amount dynamically adjusted based on health data.

[0054] The "insurance plan to be recommended" refers to the final insurance product configuration plan after dynamic adjustments, a personalized insurance plan displayed to users through the insurance recommendation system. This is used to achieve accurate insurance recommendations. For example, it can generate a "comprehensive family protection plan," including a combination recommendation of medical insurance for parents, education savings insurance for children, and property insurance linked to housing and vehicles.

[0055] Step S206: Based on the insurance plan to be recommended, recommend insurance to the user.

[0056] Insurance recommendations refer to the process of presenting recommended insurance plans to users through the insurance business system. This may include modules such as plan details display, premium calculation, and online application. For example, a user may receive a "smartly recommended family protection plan" through an insurance app, allowing them to view the terms and conditions of each product and complete the application process with a single click.

[0057] This application implements user profiles built by integrating online and offline behavioral data, breaking through the dimensional limitations of a single data source. It accurately depicts complex user characteristics and potential needs, solving the problems of incomplete coverage and coarse tagging in manual recommendations leading to inefficient matching. Secondly, the demand prediction mechanism based on a privacy-preserving computing framework achieves an intelligent upgrade in demand insight while ensuring user data security, avoiding excessive reliance on personal experience in traditional methods. The scenario interaction module dynamically demonstrates virtual insurance scenarios, not only enhancing user engagement but also capturing interactive input data in real time and dynamically adjusting insurance plans based on real-time behavioral data. The final output recommendation plan overcomes the lack of coordination caused by the fragmentation of online and offline data and, through multiple rounds of data-driven solution iteration, significantly improves the accuracy of demand matching and insurance service efficiency, thereby enhancing the accuracy of insurance recommendations.

[0058] In some optional implementations of this embodiment, step S201, obtaining the user's online behavior data and offline behavior data, specifically includes the following steps:

[0059] By using pre-set tracking technology, the user's operation path and dwell time in the application are obtained to obtain online behavior data; by using IoT devices and gateway devices, the user's behavior trajectory in physical scenarios is collected to obtain the user's offline behavior data.

[0060] Among them, event tracking technology refers to the technique of pre-setting code snippets at specific locations in an application to capture user operation information. It is used to obtain detailed behavioral data of users within the application.

[0061] The operation path represents the pages and operation steps that a user goes through from the beginning to the end of the application, and is used to analyze user behavior habits and demand preferences.

[0062] Among these metrics, dwell time is represented by data recorded using event tracking technology, indicating the user's level of attention to different pages within the application and used to assess the user's interest in the page content. For example, if a user spends 5 minutes on the details page of a critical illness insurance product in an insurance app, it reflects that they are quite interested in that product.

[0063] Among them, IoT devices are used to collect various behavioral data of users in physical scenarios. For example, smart bracelets can collect users' health data, and smart cameras can record users' activity scenarios.

[0064] Gateway devices are used to transmit data collected by IoT devices to servers. For example, smart bracelets can collect user health data and transmit it to cloud servers through gateway devices.

[0065] Among them, physical scenarios refer to the physical space environment in which users' offline behavior occurs. They come from specific physical carriers such as insurance service outlets, users' homes, and offices, and represent the real spatial dimension of users' offline service contact, which is used to supplement the scenario completeness of online data.

[0066] Among them, behavioral trajectory refers to the spatial movement path and action sequence generated by a user in a physical scenario. It originates from location coordinates, acceleration, and other data collected by IoT devices (such as in-vehicle GPS and smart bracelets). After preprocessing by gateway devices, it forms a structured record that represents the spatiotemporal dynamic characteristics of the user's offline behavior and is used to analyze the user's lifestyle habits and risk exposure patterns. For example, the driving route coordinates and the number of emergency brakings collected by in-vehicle OBD devices serve as a specific manifestation of behavioral trajectory.

[0067] In one example, using an online insurance platform, the process of acquiring user online and offline behavioral data is illustrated. Online, pre-defined tracking techniques are used to embed code snippets in key locations such as the insurance app's homepage, various insurance product detail pages, and the application process page. When a user uses the app, the tracking technique captures the user's operational path. For example, if a user clicks from the homepage to enter the accident insurance product detail page, and then enters the application process page, this path reflects the user's interest in accident insurance and potential purchase intention. Simultaneously, the time spent on each page is recorded. If a user spends 8 minutes on a life insurance product detail page, it indicates a strong interest in life insurance. Offline, IoT devices and gateway devices are used. An onboard OBD device is installed in the user's vehicle. The onboard OBD device collects data such as driving route coordinates and the number of emergency braking incidents. This data is transmitted to the server via the gateway device, forming the user's offline behavioral trajectory. For example, onboard data shows that the user frequently drives long distances at night and brakes frequently. Combined with online data, this allows for a comprehensive analysis of the user's lifestyle habits and risk exposure patterns, providing a basis for subsequent precise services.

[0068] This application's embodiments utilize event tracking technology to acquire user operation paths and dwell time within an application, thereby obtaining online behavioral data and accurately capturing detailed user behavior within the application. By recording operation paths, the complete process from user entry into the application to completion of the operation can be clearly presented, allowing for in-depth analysis of their behavioral habits and needs. Recording dwell time can assess the user's level of interest in the page content. Simultaneously, by utilizing IoT devices and gateway devices to collect user behavior trajectories in physical scenarios to obtain offline behavioral data, the completeness of online data can be supplemented, enabling comprehensive analysis of user lifestyle habits and risk exposure patterns, and providing a solid foundation for subsequent accurate construction of user profiles.

[0069] In some optional implementations, step S202 involves integrating the online behavior data and the offline behavior data using a preset multi-source data fusion engine to construct the user profile, specifically including the following steps:

[0070] The online and offline behavioral data are standardized to generate a target dataset; the target data is then processed using a preset multi-source data fusion engine and privacy protection technology to generate a user profile of the user.

[0071] The target dataset is derived from a collection of data generated after standardizing the acquired online and offline user behavior data. It represents the standardized and processed user behavior information.

[0072] Privacy protection technology refers to a series of technical measures used during data processing to ensure user privacy and security. These measures are employed to protect user privacy in areas such as multi-source data fusion and demand forecasting.

[0073] In one example, let's take an insurance company building a user profile. First, acquire online user behavior data, such as the number of times user A browses different insurance product pages on the insurance app, the duration of their browsing, and the steps they take during the insurance application process. Simultaneously, acquire offline behavior data, such as the duration of user consultations at insurance outlets and the frequency of attending offline insurance lectures. Next, standardize this online and offline behavior data. For example, convert the number of page views into a uniform value within a specific range, and standardize the duration of browsing by minute, generating a target dataset. Then, use a pre-set multi-source data fusion engine and privacy protection technologies to process the target data. For example, use differential privacy technology to encrypt sensitive information such as user identity in the dataset to prevent information leakage. After analysis by the fusion engine, it is found that user A frequently browses car insurance products online with long browsing times and frequently consults about car insurance-related issues offline. Combined with data such as their age and driving habits, a user profile is generated indicating that user A has a strong need for car insurance and pays attention to details of coverage, providing a basis for subsequent targeted marketing.

[0074] This application's embodiments generate a target dataset by standardizing online and offline behavioral data, effectively eliminating differences caused by different data sources and formats, thus ensuring data uniformity and comparability. Through a multi-source data fusion engine, privacy-preserving technologies are used to process the target data and generate user profiles. This not only integrates multi-dimensional data to comprehensively depict user characteristics and accurately grasp user insurance needs and preferences, but also protects user privacy and security, avoids the leakage of sensitive information, and enhances user trust in insurance services.

[0075] In some optional implementations, step S203, based on the user profile, generates the user's demand prediction result through a preset privacy-preserving computing framework, specifically including the following steps:

[0076] A federated learning technique using a pre-defined privacy-preserving computation framework is employed to perform cross-platform data modeling on the user profile, resulting in a comprehensive user model parameter that integrates features from multiple platforms. Using this comprehensive user model parameter as initial model parameters, and based on a pre-defined differential privacy mechanism, noise is added to the target dataset used to generate the user profile, thus generating a demand prediction model. Target behavior data of the user within a target time period is acquired. Based on feature extraction techniques and a pre-defined temporal prediction model, the target behavior data is processed to obtain behavioral temporal features. Based on these behavioral temporal features, the demand prediction model is used to generate a demand prediction result for the user.

[0077] Federated learning is a machine learning framework in which multiple participants jointly train a model through encrypted data exchange and collaborative computation without sharing the original data. It is used to achieve cross-platform and cross-institutional collaborative data modeling, avoiding the risk of data leakage.

[0078] Cross-platform data modeling refers to the process of constructing a comprehensive model using data resources from different platforms and employing unified or adapted modeling methods. It represents the ability to break down data silos and integrate features from multiple platforms to generate more comprehensive and accurate user models.

[0079] The user comprehensive model parameters are a set of numerical values ​​describing the comprehensive characteristics of a user. They represent the quantitative characteristics of a user's behavior and attributes across multiple platforms, and are used to accurately depict user profiles.

[0080] Differential privacy is a technique that protects individual privacy by adding random noise to data. It represents a method that balances privacy protection with data availability, preventing the inference of individual information from data query results.

[0081] Noise addition refers to the operation of adding random noise to the original data. It represents a method of perturbing data to protect privacy, used to hide sensitive individual information while maintaining a certain level of data usability.

[0082] The target time period is derived from the time range set for user behavior analysis; it is a pre-determined time interval used to collect and analyze user behavior data. It is used to focus on the behavioral characteristics of users within a specific period, such as analyzing a user's insurance-related behavior over the past month.

[0083] Among them, target behavior data comes from user behavior records generated within a target time period and is user behavior information related to a specific analytical purpose. It is used to analyze user needs and behavioral trends.

[0084] Feature extraction technology is a method and technique for extracting representative features from raw data. Characterization is the process of transforming raw data into features that can be used for modeling and analysis, thereby reducing data dimensionality and improving model performance.

[0085] Among them, the time-series forecasting model is a mathematical model that predicts future values ​​based on historical time-series data. It is used to predict future user behavior and needs.

[0086] Among them, behavioral temporal features are features extracted after performing temporal analysis on target behavioral data, and are a set of features reflecting the changing patterns of user behavior over time. These dynamic temporal features characterize user behavior and are used to more accurately predict user needs.

[0087] In one example, a large insurance group with multiple business segments including property and casualty insurance and life insurance, and partnerships with banks and other financial institutions, possesses user data from multiple platforms. First, federated learning technology using a pre-defined privacy-preserving computational framework is employed to perform cross-platform data modeling of user profiles built on each platform. Each platform trains its sub-model locally using user data, exchanging only model parameters, not raw data. For instance, the sub-model trained by the property and casualty insurance platform based on user car insurance purchase records interacts with the sub-model trained by the life insurance platform based on user health insurance data, ultimately yielding a comprehensive user model parameter that integrates features from multiple platforms, fully reflecting users' various preferences and risk characteristics in the insurance field. Next, using the comprehensive user model parameter as the initial model parameter, noise is added to the target dataset used to generate the user profiles based on a pre-defined differential privacy mechanism. For example, random noise is added to user age data, ensuring that the data can still be used for modeling while protecting user privacy, thereby generating a demand prediction model. Then, target behavior data of users over the past three months is obtained, such as the number of times users browse insurance product pages and the frequency of contacting customer service. Based on feature extraction technology, key features are extracted from this data. Then, a pre-defined time-series prediction model is used to analyze the changing patterns of these features over time, yielding behavioral time-series features. Finally, based on these behavioral time-series features, a demand prediction model is employed to predict the user's future demand for different types of insurance products. For example, if the prediction indicates a high demand for short-term accident insurance, this provides a basis for subsequent accurate insurance product recommendations.

[0088] This application's embodiments utilize federated learning technology within a privacy-preserving computation framework for cross-platform data modeling. This allows for the integration of user profile information from multiple platforms without sharing raw data, fully mining multi-dimensional user characteristics to obtain comprehensive user model parameters that fuse multi-platform features, thereby enhancing data utilization value and security. Using these parameters as initial parameters, a demand prediction model is generated by adding noise to the target dataset using a differential privacy mechanism, protecting user privacy while ensuring model accuracy. Target behavior data within a target time period is acquired, and through feature extraction and temporal prediction model processing, behavioral temporal features are obtained, capturing dynamic changes in user behavior. Finally, based on these features, the demand prediction model generates results that accurately predict user insurance needs, providing strong support for personalized recommendations.

[0089] In some optional implementations, step S204, based on the demand prediction results, uses a preset scene interaction module to display dynamic demonstration content of a virtual insurance scenario to the user in order to obtain the user's interactive input data, specifically including the following steps:

[0090] Based on the demand prediction results, a dynamic demonstration of a virtual insurance scenario is generated through a preset scenario interaction module; the dynamic demonstration is displayed to the user; and interactive input data generated by the user's interaction with the dynamic demonstration is obtained through preset gesture recognition and voice recognition technologies.

[0091] Gesture recognition technology is a technique that uses image acquisition devices to acquire user gesture images or video streams, uses algorithms to analyze and process them, extracts gesture features, and identifies the specific gesture type.

[0092] Speech recognition technology refers to the technology of collecting users' voice signals using devices such as microphones, converting them into electrical signals, and then analyzing and decoding the signals using acoustic and language models to convert them into text or commands. For example, it can recognize users' spoken insurance-related questions or requests.

[0093] In one example, taking an online insurance platform, the platform first generates demand predictions based on user profiles, such as predicting a user's high demand for short-term travel insurance. Next, through a pre-defined scenario interaction module, combined with information such as travel destination and mode of transportation, it generates dynamic demonstrations of virtual travel insurance scenarios, such as simulating the medical treatment and claims process for an accidental injury at a tourist attraction. After showing this dynamic demonstration to the user, the user interacts with the content through gestures and voice. For example, the user waves to switch between different insurance terms and speaks voice commands such as "Can this deductible be reduced?" At this time, pre-defined gesture recognition technology captures the user's gestures and converts them into operation commands, while voice recognition technology converts the user's speech into text information. Both technologies together obtain the interactive input data generated by the user's interaction with the dynamic demonstration.

[0094] This application embodiment generates dynamic demonstration content of virtual insurance scenarios based on demand prediction results and displays it to users using a scene interaction module, breaking the traditional static mode of insurance presentation. The dynamic demonstration content presents insurance-related information in a more intuitive and vivid way, allowing users to more clearly understand the features and coverage of insurance products. Simultaneously, by utilizing preset gesture recognition and voice recognition technologies to obtain user interaction input data with the dynamic demonstration content, natural interaction between users and the virtual scene is achieved, greatly enhancing user participation and experience, and enabling users to more proactively explore insurance options.

[0095] In some optional implementations, after the step "obtaining the interactive input data generated by the user's interaction with the dynamic demonstration content through preset gesture recognition and voice recognition technologies", the following steps are also included:

[0096] Based on preset natural language processing technology, the interactive input data is processed to generate a virtual object that responds to the user in real time.

[0097] Natural Language Processing (NLP) technology, originating from the intersection of artificial intelligence and linguistics, is a collection of algorithms and techniques that enable computers to understand, analyze, and generate human natural language. It is used to analyze the intent and emotion in user interaction input data. For example, it can extract key information and the user's intended need when analyzing a user's voice inquiry about insurance terms.

[0098] Real-time response refers to the process by which the system generates and feeds back corresponding results based on preset rules or algorithms after receiving user interaction input data. For example, after a user issues a gesture command, the virtual object immediately responds with the corresponding action.

[0099] In one example, in the aforementioned internet insurance platform case, after acquiring the user's interactive input data with the dynamic demonstration content of the virtual insurance scenario, it can be processed based on preset natural language processing (NLP) technology. For instance, if a user asks, "Does this travel insurance cover flight delay compensation?", the NLP technology first performs preprocessing on the text converted from speech, such as word segmentation and part-of-speech tagging. Then, through semantic analysis, it understands that the user's intention is to inquire about flight delay compensation within the insurance coverage. Next, based on preset rules and a knowledge base, the system generates a virtual object (such as a virtual insurance advisor) to provide a real-time response to the target object (user), such as, "This travel insurance includes flight delay compensation; a certain amount of compensation is available for delays of 4 hours or more." Through NLP technology, intelligent interaction between the virtual object and the user is achieved, promptly answering user questions and improving user experience and purchase intention.

[0100] This application's embodiments utilize natural language processing (NLP) technology to process interactive input data, thereby generating a virtual object's real-time response to the target object, offering significant advantages. NLP technology can accurately interpret the complex intentions and needs of users through gestures and voice input, transforming ambiguous expressions into clear instructions. The virtual object then responds in real-time, creating a scenario similar to interacting with a real customer service representative, enhancing the interactivity and realism between the user and the system. This real-time, intelligent response method not only promptly answers user questions but also guides users to further understand insurance products based on feedback, effectively increasing user acceptance of insurance plans.

[0101] In some optional implementations, real-time behavioral data includes health monitoring data and driving behavior data. Step S205 involves adjusting the user's insurance plan based on the user profile, the interaction input data, and the real-time behavioral data to obtain a recommended insurance plan. This specifically includes the following steps:

[0102] The user profile data, interactive input data, and real-time behavior data are integrated to obtain a standardized data set. A preset reinforcement learning algorithm is used to analyze the data in the standardized data set to obtain a behavior decision set. If the behavior decision set meets a preset dynamic adjustment threshold, the user's insurance plan is adjusted to obtain a recommended insurance plan.

[0103] The user profile data is derived from user profiles constructed by integrating online and offline behavioral data through a multi-source data fusion engine. It represents multi-dimensional user characteristics, such as consumption habits and interests.

[0104] The standardized dataset is comprised of user profile data, interactive input data, and real-time behavioral data (such as health monitoring data and driving behavior data). It represents a dataset that has undergone standardized and uniform processing.

[0105] Reinforcement learning algorithms are a type of algorithm that learns optimal behavioral strategies by having an agent interact with its environment and learn from the rewards or punishments received. It represents the ability to continuously optimize decisions based on data feedback.

[0106] The behavioral decision set is the result of analyzing a standardized dataset using a reinforcement learning algorithm. It represents a series of possible behavioral decisions generated based on the user's situation. These decisions are used to determine whether to adjust the user's insurance plan. For example, it may include decision information such as increasing the coverage amount or changing the type of insurance.

[0107] The dynamic adjustment threshold is a pre-defined numerical range used to determine whether the behavioral decision set meets the conditions for adjusting the insurance plan. It represents the critical criterion for determining whether to dynamically adjust the insurance plan.

[0108] In one example, taking an online auto insurance platform, the platform first acquires user B's online behavioral data (such as browsing history and consultation records on insurance-related websites) and offline behavioral data (such as the number of times they consulted with offline stores). Using a multi-source data fusion engine, a user profile of user B is constructed, showing that they are sensitive to auto insurance prices and focus on coverage. Based on this user profile, a privacy-preserving computational framework predicts that user B has a need to reduce premiums while increasing specific coverage. The scenario interaction module displays a dynamic demonstration of a virtual auto insurance scenario. User B asks via voice, "How much will adding flood insurance increase the premium?" This is the interactive input data. Simultaneously, the platform acquires user B's real-time behavioral data, including health monitoring data (showing their recent good health) and driving behavior data (showing their recent stable driving with no violations). Next, the profile data, interactive input data, and real-time behavioral data are integrated, unifying the data format and dimensions to obtain a standardized dataset. Using a pre-set reinforcement learning algorithm, with the optimization objective of minimizing premiums and maximizing coverage, the standardized dataset is analyzed to obtain a behavioral decision set, such as suggesting adding flood insurance or adjusting the deductible. The preset dynamic adjustment threshold is that the premium change does not exceed 15% of the original premium and the coverage is expanded. This action's decision set meets this threshold, so the platform adjusts user B's insurance plan, adding flood insurance and appropriately increasing the deductible, resulting in a recommended insurance plan, which is then recommended to user B, achieving accurate insurance recommendation.

[0109] This application embodiment integrates user profile data, interactive input data, and real-time behavioral data including health monitoring and driving behavior data to obtain a standardized data set, breaking down barriers between different data types. Reinforcement learning algorithms are used to analyze the standardized data set, enabling it to learn and make decisions autonomously based on dynamic data changes, generating a behavioral decision set tailored to the user's actual needs. When the behavioral decision set meets a preset dynamic adjustment threshold, the insurance plan is adjusted, ensuring the rationality and scientific nature of the adjustment, avoiding blind changes, and ultimately resulting in a more accurate insurance plan to match the user, improving the accuracy of insurance recommendations and user acceptance.

[0110] It should be emphasized that, in order to further ensure the privacy and security of the aforementioned online behavior data, offline behavior data, user profiles, demand prediction results, interactive input data, and insurance plans to be recommended, these data can also be stored in a blockchain node.

[0111] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0112] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware with computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0113] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0114] Further reference Figure 3 As a response to the above Figure 2 The implementation of the method shown in this application provides an embodiment of an insurance recommendation device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0115] like Figure 3 As shown, the insurance recommendation device 400 of this embodiment includes: an acquisition module 401, an integration module 402, a generation module 403, a display module 404, an adjustment module 405, and a recommendation module 406. Wherein:

[0116] Module 401 is used to acquire users' online and offline behavior data.

[0117] The integration module 402 is used to integrate the online behavior data and the offline behavior data through a preset multi-source data fusion engine to construct the user profile of the user.

[0118] The generation module 403 is used to generate the user's demand prediction result based on the user profile and through a preset privacy-preserving computing framework;

[0119] The display module 404 is used to display dynamic demonstration content of a virtual insurance scenario to the user through a preset scene interaction module based on the demand prediction results, so as to obtain the user's interactive input data;

[0120] The adjustment module 405 is used to acquire the user's real-time behavior data, and adjust the user's insurance plan based on the user profile, the interaction input data, and the real-time behavior data to obtain the insurance plan to be recommended.

[0121] The recommendation module 406 is used to recommend insurance to the user based on the insurance plan to be recommended.

[0122] This embodiment overcomes the dimensional limitations of a single data source by constructing user profiles through the integration of online and offline behavioral data. It accurately depicts complex user characteristics and potential needs, resolving the inefficiencies in matching caused by incomplete coverage and coarse tagging in manual recommendations. Secondly, the demand prediction mechanism based on a privacy-preserving computing framework achieves an intelligent upgrade in demand insight while ensuring user data security, avoiding excessive reliance on personal experience in traditional methods. The scenario interaction module dynamically demonstrates virtual insurance scenarios, enhancing user engagement and capturing interactive input data in real time, dynamically adjusting insurance plans based on real-time behavioral data. The final recommended plan overcomes the lack of coordination caused by the fragmentation of online and offline data, and through multiple rounds of data-driven plan iteration, significantly improves the accuracy of demand matching and insurance service efficiency, thereby enhancing the accuracy of insurance recommendations.

[0123] In one embodiment, the acquisition module 401 includes:

[0124] The first acquisition submodule is used to obtain the user's operation path and dwell time in the application through preset tracking technology, and obtain online behavior data.

[0125] The data acquisition submodule is used to collect the user's behavioral trajectory in a physical scene through IoT devices and gateway devices to obtain the user's offline behavioral data.

[0126] This application's embodiments utilize event tracking technology to acquire user operation paths and dwell time within an application, thereby obtaining online behavioral data and accurately capturing detailed user behavior within the application. By recording operation paths, the complete process from user entry into the application to completion of the operation can be clearly presented, allowing for in-depth analysis of their behavioral habits and needs. Recording dwell time can assess the user's level of interest in the page content. Simultaneously, by utilizing IoT devices and gateway devices to collect user behavior trajectories in physical scenarios to obtain offline behavioral data, the completeness of online data can be supplemented, enabling comprehensive analysis of user lifestyle habits and risk exposure patterns, and providing a solid foundation for subsequent accurate construction of user profiles.

[0127] In one embodiment, the integration module 402 includes:

[0128] The standardization submodule is used to standardize the online and offline behavioral data to generate a target dataset;

[0129] The first processing submodule is used to process the target data using a preset multi-source data fusion engine and privacy protection technology to generate a user profile of the user.

[0130] This application's embodiments generate a target dataset by standardizing online and offline behavioral data, effectively eliminating differences caused by different data sources and formats, thus ensuring data uniformity and comparability. Through a multi-source data fusion engine, privacy-preserving technologies are used to process the target data and generate user profiles. This not only integrates multi-dimensional data to comprehensively depict user characteristics and accurately grasp user insurance needs and preferences, but also protects user privacy and security, avoids the leakage of sensitive information, and enhances user trust in insurance services.

[0131] In one embodiment, the generation module 403 includes:

[0132] The modeling submodule is used to perform cross-platform data modeling on the user profile using federated learning technology with a preset privacy-preserving computing framework, and to obtain comprehensive user model parameters that integrate features from multiple platforms.

[0133] The noise-adding submodule is used to add noise to the target dataset for generating the user profile based on the user comprehensive model parameters as the initial parameters of the model and a preset differential privacy mechanism to generate a demand prediction model.

[0134] The second acquisition submodule is used to acquire the user's target behavior data within the target time period;

[0135] The second processing submodule is used to process the target behavior data based on feature extraction technology and a preset temporal prediction model to obtain behavioral temporal features;

[0136] The result generation submodule is used to generate the user's demand prediction results based on the behavioral time-series characteristics and the demand prediction model.

[0137] This application's embodiments utilize federated learning technology within a privacy-preserving computation framework for cross-platform data modeling. This allows for the integration of user profile information from multiple platforms without sharing raw data, fully mining multi-dimensional user characteristics to obtain comprehensive user model parameters that fuse multi-platform features, thereby enhancing data utilization value and security. Using these parameters as initial parameters, a demand prediction model is generated by adding noise to the target dataset using a differential privacy mechanism, protecting user privacy while ensuring model accuracy. Target behavior data within a target time period is acquired, and through feature extraction and temporal prediction model processing, behavioral temporal features are obtained, capturing dynamic changes in user behavior. Finally, based on these features, the demand prediction model generates results that accurately predict user insurance needs, providing strong support for personalized recommendations.

[0138] In one embodiment, the display module 404 includes:

[0139] The content generation submodule is used to generate dynamic demonstration content of virtual insurance scenarios based on the demand prediction results and through a preset scene interaction module.

[0140] The display submodule is used to display the dynamic demonstration content to the user;

[0141] The third acquisition submodule is used to acquire interactive input data generated by the user's interaction with the dynamic demonstration content through preset gesture recognition and voice recognition technologies.

[0142] This application embodiment generates dynamic demonstration content of virtual insurance scenarios based on demand prediction results and displays it to users using a scene interaction module, breaking the traditional static mode of insurance presentation. The dynamic demonstration content presents insurance-related information in a more intuitive and vivid way, allowing users to more clearly understand the features and coverage of insurance products. Simultaneously, by utilizing preset gesture recognition and voice recognition technologies to obtain user interaction input data with the dynamic demonstration content, natural interaction between users and the virtual scene is achieved, greatly enhancing user participation and experience, and enabling users to more proactively explore insurance options.

[0143] In one embodiment, the third acquisition submodule is further configured to process the interactive input data based on a preset natural language processing technology to generate a virtual object's real-time response to the user.

[0144] This application's embodiments utilize natural language processing (NLP) technology to process interactive input data, thereby generating a virtual object's real-time response to the target object, offering significant advantages. NLP technology can accurately interpret the complex intentions and needs of users through gestures and voice input, transforming ambiguous expressions into clear instructions. The virtual object then responds in real-time, creating a scenario similar to interacting with a real customer service representative, enhancing the interactivity and realism between the user and the system. This real-time, intelligent response method not only promptly answers user questions but also guides users to further understand insurance products based on feedback, effectively increasing user acceptance of insurance plans.

[0145] In one embodiment, the adjustment module 405 includes:

[0146] The integration submodule is used to integrate the user profile data, the interactive input data, and the real-time behavior data to obtain a standardized data set;

[0147] The analysis submodule is used to analyze the data in the standardized dataset using a preset reinforcement learning algorithm to obtain a set of behavioral decisions.

[0148] The adjustment submodule is used to adjust the user's insurance plan if the behavior decision set meets the preset dynamic adjustment threshold, so as to obtain the insurance plan to be recommended.

[0149] This application embodiment integrates user profile data, interactive input data, and real-time behavioral data including health monitoring and driving behavior data to obtain a standardized data set, breaking down barriers between different data types. Reinforcement learning algorithms are used to analyze the standardized data set, enabling it to learn and make decisions autonomously based on dynamic data changes, generating a behavioral decision set tailored to the user's actual needs. When the behavioral decision set meets a preset dynamic adjustment threshold, the insurance plan is adjusted, ensuring the rationality and scientific nature of the adjustment, avoiding blind changes, and ultimately resulting in a more accurate insurance plan to match the user, improving the accuracy of insurance recommendations and user acceptance.

[0150] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0151] Computer device 6 includes a memory 61, a processor 62, and a network interface 63 that are interconnected via a system bus. It should be noted that only computer device 6 with memory 61, processor 62, and network interface 63 is shown in the figure; however, it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0152] Computer devices can include desktop computers, laptops, handheld computers, and cloud servers. These devices allow for human-computer interaction with users through methods such as keyboards, mice, remote controls, touchpads, or voice-activated devices.

[0153] The memory 61 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 61 may be an internal storage unit of the computer device 6, such as the hard disk or memory of the computer device 6. In other embodiments, the memory 61 may also be an external storage device of the computer device 6, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 6. Of course, the memory 61 may include both the internal storage unit and the external storage device of the computer device 6. In this embodiment, the memory 61 is typically used to store the operating system and various application software installed on the computer device 6, such as computer-readable instructions for insurance recommendation methods. In addition, the memory 61 may also be used to temporarily store various types of data that have been output or will be output.

[0154] In some embodiments, processor 62 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. Processor 62 is typically used to control the overall operation of computer device 6. In this embodiment, processor 62 is used to execute computer-readable instructions stored in memory 61 or to process data, such as computer-readable instructions for executing insurance recommendation methods.

[0155] The network interface 63 may include a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the computer device 6 and other electronic devices.

[0156] This application's embodiments, by integrating online and offline behavioral data to construct user profiles, overcome the dimensional limitations of single data sources, accurately depicting complex user characteristics and potential needs, and solving the problems of incomplete coverage and coarse tagging in manual recommendations leading to inefficient matching. Secondly, the demand prediction mechanism based on a privacy-preserving computing framework achieves an intelligent upgrade in demand insight while ensuring user data security, avoiding excessive reliance on personal experience in traditional methods. The scenario interaction module, through dynamically demonstrating virtual insurance scenarios, not only enhances user engagement but also captures interactive input data in real time, dynamically adjusting insurance plans based on real-time behavioral data. The final output recommendation plan overcomes the lack of coordination caused by the fragmentation of online and offline data, and through multiple rounds of data-driven plan iteration, significantly improves the accuracy of demand matching and insurance service efficiency, thereby enhancing the accuracy of insurance recommendations.

[0157] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the insurance recommendation method described above.

[0158] This application's embodiments, by integrating online and offline behavioral data to construct user profiles, overcome the dimensional limitations of single data sources, accurately depicting complex user characteristics and potential needs, and solving the problems of incomplete coverage and coarse tagging in manual recommendations leading to inefficient matching. Secondly, the demand prediction mechanism based on a privacy-preserving computing framework achieves an intelligent upgrade in demand insight while ensuring user data security, avoiding excessive reliance on personal experience in traditional methods. The scenario interaction module, through dynamically demonstrating virtual insurance scenarios, not only enhances user engagement but also captures interactive input data in real time, dynamically adjusting insurance plans based on real-time behavioral data. The final output recommendation plan overcomes the lack of coordination caused by the fragmentation of online and offline data, and through multiple rounds of data-driven plan iteration, significantly improves the accuracy of demand matching and insurance service efficiency, thereby enhancing the accuracy of insurance recommendations.

[0159] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods of the various embodiments of this application.

[0160] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

[0161] The software tools or components not belonging to our company that appear in the embodiments of this application are merely examples and do not represent actual use.

Claims

1. An insurance recommendation method, characterized in that, Includes the following steps: Acquire users' online and offline behavior data; By using a pre-set multi-source data fusion engine, the online behavior data and the offline behavior data are integrated to construct the user profile of the user; Based on the user profile, a pre-defined privacy-preserving computing framework is used to generate a prediction result of the user's needs. Based on the demand forecast results, a dynamic demonstration of a virtual insurance scenario is presented to the user through a preset scenario interaction module in order to obtain the user's interactive input data. The user's real-time behavior data is obtained, and the user's insurance plan is adjusted based on the user profile, the interaction input data, and the real-time behavior data to obtain the insurance plan to be recommended. Based on the insurance plan to be recommended, insurance recommendations are made to the user.

2. The method according to claim 1, characterized in that, The steps for obtaining users' online and offline behavior data specifically include: By using pre-set tracking technology, we can obtain the user's operation path and dwell time in the application and obtain online behavior data. By using IoT devices and gateway devices, the user's behavioral trajectory in physical scenarios is collected to obtain the user's offline behavioral data.

3. The method according to claim 1, characterized in that, The step of integrating the online behavior data and the offline behavior data through a preset multi-source data fusion engine to construct the user profile specifically includes: The online and offline behavioral data are standardized to generate a target dataset; The target data is processed using a pre-set multi-source data fusion engine and privacy protection technology to generate a user profile for the user.

4. The method according to claim 3, characterized in that, The step of generating the user's demand prediction result based on the user profile and through a preset privacy-preserving computing framework specifically includes: Federated learning technology using a pre-defined privacy-preserving computing framework is used to perform cross-platform data modeling on the user profile, resulting in comprehensive user model parameters that integrate features from multiple platforms. Using the user comprehensive model parameters as initial model parameters, and based on a preset differential privacy mechanism, the target dataset for generating the user profile is noise-added to generate a demand prediction model. Obtain the target behavior data of the user within the target time period; Based on feature extraction technology and a preset temporal prediction model, the target behavior data is processed to obtain behavioral temporal features; Based on the aforementioned behavioral time-series characteristics, the aforementioned demand prediction model is used to generate the user's demand prediction results.

5. The method according to claim 1, characterized in that, The step of displaying a dynamic demonstration of a virtual insurance scenario to the user based on the demand prediction results, through a preset scenario interaction module, in order to obtain the user's interactive input data, specifically includes: Based on the demand forecast results, dynamic demonstration content of the virtual insurance scenario is generated through a preset scenario interaction module. The dynamic demo content is displayed to the user. The system uses preset gesture recognition and voice recognition technologies to obtain interactive input data generated by the user's interaction with the dynamic demonstration content.

6. The method according to claim 5, characterized in that, After the step of acquiring the interactive input data generated by the user's interaction with the dynamic demonstration content through preset gesture recognition and voice recognition technologies, the method further includes: Based on preset natural language processing technology, the interactive input data is processed to generate a virtual object that responds to the user in real time.

7. The method according to claim 1, characterized in that, The real-time behavioral data includes health monitoring data and driving behavior data; the step of adjusting the user's insurance plan based on the user profile, the interactive input data, and the real-time behavioral data to obtain a recommended insurance plan specifically includes: The user profile data, the interactive input data, and the real-time behavior data are integrated to obtain a standardized data set. A preset reinforcement learning algorithm is used to analyze the data in the standardized dataset to obtain a behavioral decision set; If the behavioral decision set meets the preset dynamic adjustment threshold, then the user's insurance plan is adjusted to obtain the insurance plan to be recommended.

8. An insurance recommendation device, characterized in that, include: The acquisition module is used to acquire users' online and offline behavior data; The integration module is used to integrate the online behavior data and the offline behavior data through a preset multi-source data fusion engine to construct the user profile of the user; The generation module is used to generate the user's demand prediction results based on the user profile and through a preset privacy-preserving computing framework. The display module is used to present dynamic demonstration content of a virtual insurance scenario to the user through a preset scene interaction module based on the demand prediction results, so as to obtain the user's interactive input data. The adjustment module is used to acquire the user's real-time behavior data, and adjust the user's insurance plan based on the user profile, the interaction input data, and the real-time behavior data to obtain the insurance plan to be recommended. The recommendation module is used to recommend insurance to the user based on the insurance plan to be recommended.

9. A computer device, characterized in that, The system includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the insurance recommendation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions that, when executed by a processor, implement the steps of the insurance recommendation method as described in any one of claims 1 to 7.

Images