Insurance product customization system and method based on natural language intent analysis

Abstract

The application relates to the technical field of natural language processing, and discloses an insurance product customization system and method based on natural language intention analysis, which can be applied to an insurance business scene. The system can acquire user demand information and user portrait data, use a demand disassembly intelligent agent to disassemble core elements from the user demand information, generate standard demand tags according to the core elements, generate additional demand tags according to the standard demand tags and the user portrait data, use a product assembly planning intelligent agent group to generate a product customization result based on the standard demand tags and the additional demand tags, and then display the product customization result. The system can independently analyze user colloquial demands in steps through the demand disassembly intelligent agent, automatically assemble products through the component planning intelligent agent, and generate material combinations through a material planning intelligent agent, so that the richness and accuracy of information related to insurance products can be improved.

Description

Technical Field

[0001] This application relates to the field of natural language processing technology, and in particular to an insurance product customization system and method based on natural language intent parsing. Background Technology

[0002] An insurance product customization system is a visual information interaction system. It can be configured on client devices such as mobile terminals and computers to display customized interfaces related to insurance products. Based on user interactions within these interfaces, the system generates, combines, processes, and displays relevant insurance product information.

[0003] For example, an insurance product customization system can display a customization interface for insurance products. This interface can include standardized form-related customization options such as main insurance, supplementary insurance, adjusting the sum insured, and configuring value-added services. Users can select one or more customization options from the interface according to their needs. The insurance product customization system then assembles the relevant insurance products based on the user's selected options and generates product customization information based on the assembly results for visual display.

[0004] However, because the insurance product customization system only supports selecting customization options on standardized forms, the entire process relies on manual operation. Furthermore, the customization options and generated insurance product information are pre-edited manually, lacking the ability to autonomously plan generation logic and dynamically adjust solutions. When using technologies like large language models for insurance product customization, the system can only match semantically similar product combinations from a pre-defined product library based on the model's semantic parsing results. This results in product combinations that are only semantically relevant to the user's input, ignoring the user's implicit needs. Consequently, the richness and accuracy of the insurance product information generated by the insurance product customization system are low. Summary of the Invention

[0005] In view of this, embodiments of this application provide an insurance product customization system and method based on natural language intent parsing to solve the problem of low richness and accuracy of insurance product-related information generated by the insurance product customization system.

[0006] According to a first aspect of this application, an insurance product customization system based on natural language intent parsing is provided, the system comprising: The client is configured to perform user interactions, receiving user input and displaying product customization results; The server establishes a communication connection with the client. The server is configured with a group of intelligent agents for demand breakdown and product assembly planning. The product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent. The server is configured as follows: Obtain user demand information and user profile data of the target users corresponding to the user demand information; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The standard requirement tags and the additional requirement tags are input into the product assembly planning intelligent agent group to obtain the product customization results output by the product assembly planning intelligent agent group. The product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent. The product customization results are sent to the client so that the product customization results can be displayed through the client.

[0007] According to a second aspect of this application, a method for customizing insurance products based on natural language intent parsing is provided, the method comprising: Obtain user demand information and user profile data of the target users corresponding to the user demand information; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The standard requirement tags and the additional requirement tags are input into the product assembly planning intelligent agent group to obtain the product customization results output by the product assembly planning intelligent agent group; the product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent; the product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent; The product customization results are sent to the client so that the product customization results can be displayed through the client.

[0008] According to a third aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the above-described method for customizing insurance products based on natural language intent parsing.

[0009] According to a fourth aspect of this application, a storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the above-described method for customizing insurance products based on natural language intent parsing.

[0010] In the aforementioned insurance product customization system based on natural language intent parsing, the server can obtain user demand information and target user profile data from the client. A demand decomposition agent then extracts core elements from the user demand information and generates standard demand tags based on these core elements. Additional demand tags are then generated based on the standard demand tags and user profile data. Finally, a product assembly planning agent group generates the product customization result based on the standard and additional demand tags, and sends the product customization result to the client for display. This system can autonomously and step-by-step parse users' spoken demands through the demand decomposition agent, automatically assemble products through the component planning agent, and generate material combinations through the supporting material planning agent, thereby improving the richness and accuracy of insurance product-related information.

[0011] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of the insurance product customization system structure provided in the embodiments of this application; Figure 2 This is a schematic diagram of the insurance product customization method provided in the embodiments of this application; Figure 3 This is a schematic diagram of the intelligent agent processing flow provided in the embodiments of this application; Figure 4 This is a schematic diagram of the secondary custom adjustment guarantee responsibility process provided in the embodiments of this application; Figure 5 This is a schematic diagram of the processing flow of the insurance product customization system provided in the embodiments of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] In this embodiment, insurance product customization is an information generation process in application scenarios such as finance and medical insurance, which can generate personalized insurance product information based on user needs. Here, user needs refer to the set of required items or semantics determined based on user-input interactive operations or text data. Insurance product information refers to recommendation information containing at least one insurance product that can be presented to the user.

[0016] To generate personalized insurance product information, an insurance product customization system can be built specifically for insurance product customization services. This insurance product customization system is a visual information interaction system. For example... Figure 1 As shown, the insurance product customization system can be configured as a system consisting of a client and a server. The client can include devices such as mobile terminals and computers, used to display the customized interface related to the insurance product and perform user interactions. The client can also include applications, web pages, and operating systems deployed on the terminal devices.

[0017] Depending on the business scenario corresponding to the application environment, clients can be of different types. For example, in a financial insurance scenario, clients can be categorized into user clients, business clients, and operations and maintenance (O&M) clients based on the user's target audience. User clients can interact with users on their own devices through applications, mini-programs, or web pages. Business clients can interact with business personnel on their own devices through applications, mini-programs, or web pages. O&M clients can interact with O&M personnel on their own devices through a backend management system (web) or monitoring equipment.

[0018] The server can be deployed in the cloud or data center, responsible for core business processing, persistent data storage, security management, and integration with external systems. Based on user interactions through a customized interface, the server can generate, combine, and process insurance product information, thus creating insurance product details. This information is then sent to the client for display through their customized interface.

[0019] In some embodiments, the server may include one or more combinations of servers such as web servers, application servers, database servers, and storage servers. Furthermore, depending on the type of server, corresponding functional components may be deployed within it. For example, the server may include a memory and a processor, wherein the memory stores applications related to insurance product customization and data related to insurance products for execution and retrieval by the processor. The processor executes the applications related to insurance product customization to implement the insurance product customization function.

[0020] The client and server can establish a communication connection via a network, thereby building an insurance product customization system. For example, the system can display a customization interface for insurance products on the client side. This interface can include standardized form options for customization such as main insurance, supplementary insurance, adjusting the sum insured, and configuring value-added services. Users can select one or more customization options according to their needs. The client can detect the user's selection action, generate user input information, and send it to the server. The server can then extract the corresponding insurance product data based on the selected options, assemble the relevant insurance products, and generate product customization information based on the assembly result. This customized product information is then sent to the client for visual display.

[0021] In some embodiments, the insurance product customization system can be integrated into other forms of interactive platforms as a functional module of the interactive platform. For example, the insurance product customization system can be integrated into an intelligent customer service platform. When a user uses the intelligent customer service platform to input natural language text with the intent related to insurance product recommendations for intelligent question answering, the insurance product customization system can be invoked, and an answer containing personalized insurance product information can be generated based on the insurance product customization system and displayed to the user.

[0022] When a user enters a personalized insurance request through the intelligent customer service platform, such as "to purchase low-cost medical insurance for family members, while also covering travel accidents," the intelligent customer service platform can call a Large Language Model (LLM) to perform semantic recognition on the input request text. Based on the semantic recognition results, it matches relevant knowledge items of insurance products, assembles and generates insurance product information, and then uses natural language generation technology to transform the knowledge items and insurance product information into response text, which is then pushed to the client for display.

[0023] However, because the insurance product customization system only supports selecting customization options on standardized forms, the entire process relies on manual operation. Furthermore, the customization options and generated insurance product information are pre-edited manually, lacking the ability to autonomously plan generation logic and dynamically adjust solutions. When users input text describing their personalized protection needs, the static large model passively outputs only the required fields, unable to break down core elements such as coverage, sum insured, service period, and supplementary insurance. This requires manual customer service to manually input and break down the requirements line by line, resulting in low efficiency. Consequently, the richness and accuracy of the insurance product information generated by the insurance product customization system are low.

[0024] Meanwhile, the assembly process for the aforementioned insurance products is cumbersome and lacks automation and autonomous planning. It requires manual selection of main and supplementary insurance, adjustment of coverage amounts, and configuration of value-added services. The entire product generation process relies on manual operation, lacking an agentic intelligent agent with autonomous decision-making capabilities and initiative to autonomously plan component combinations. It cannot automatically complete the flexible combination of liabilities and customized product assembly based on real-time customer needs and intentions. Furthermore, the generation of marketing materials depends on manual editing, lacking autonomous planning of copywriting and graphic material generation logic. Each customized product requires manual production of introductory materials, resulting in long delivery cycles.

[0025] Furthermore, the personalized product recommendations provided by the aforementioned insurance product customization system lack accuracy. Because the recommendation rules rely solely on fixed audience tags to push standardized packages, they lack the ability to dynamically plan and adjust solutions. They cannot autonomously plan personalized recommendation paths based on real-time customer dialogue intent and dynamically segmented customer profiles. The high barrier to entry for custom product creation at the team level makes it difficult to quickly and flexibly combine coverage responsibilities according to customers' diverse needs. Therefore, the personalized customization experience for customers is poor, resulting in low conversion rates for final solutions.

[0026] In some embodiments, insurance product customization can also be achieved using technologies such as large language models. This involves inputting the user's input request text into a large language model to obtain the semantic analysis results output by the model. Then, based on the semantic analysis results, products are matched against a pre-defined product database to obtain product combinations with high semantic similarity.

[0027] However, since the semantic matching process is based on indicators such as cosine similarity, the resulting product combinations are only semantically related to the user input, ignoring the user's implicit needs. Furthermore, simple semantic matching cannot achieve autonomous planning of product customization tasks, resulting in low richness and accuracy of insurance product information generated by the insurance product customization system.

[0028] To address the issue of low richness and accuracy of insurance product information generated by insurance product customization systems, this application provides a method for insurance product customization based on natural language intent parsing in some embodiments. This method can autonomously and step-by-step parse users' spoken language needs through a demand decomposition agent, automatically assemble products through a component planning agent, and generate material combinations through a matching material planning agent, thereby improving the richness and accuracy of insurance product information.

[0029] The method can be applied to an insurance product customization system based on natural language intent parsing. The method can also be applied to electronic devices that establish a communication connection with the insurance product customization system and have data processing capabilities. These electronic devices include, but are not limited to, computers, servers, mobile terminals, smart wearable devices, and industrial control machines. For ease of description, in this embodiment, the server in the insurance product customization system is used as the execution subject of the method. It should be understood that the method can also be applied to other types of execution subjects, which are not shown one by one in this embodiment. Figure 2 As shown, the method includes: S101. Obtain user demand information and user profile data of the target users corresponding to the user demand information.

[0030] During the insurance product customization process, the server can obtain user requirement information through the client. This user requirement information includes natural language text input from multiple sources. Therefore, the client in the insurance product customization system can be configured to receive user input information and send the received information to the server.

[0031] For example, the insurance product customization system can support multi-source natural language access. By connecting the server to a unified omnichannel information input platform, it can receive natural language text such as user text inquiries, customer service voice-to-text transcription, and salesperson input of short requirement sentences through the client. The natural language text can be used to freely describe the target user's insurance needs.

[0032] After obtaining user demand information, user profile data corresponding to the target users can be obtained based on this information. This user profile data comprises all raw data and their derived tags collected, processed, and used to depict user characteristics for constructing user profiles; it is a structured data set obtained after highly refining user information.

[0033] For example, the insurance product customization system can be integrated with or connected to a user profiling system. After user authorization, the user profiling system can collect anonymized user behavior data from multiple sources, and then perform data cleaning, ID integration, and tag generation on this data to create user profile data. This user profile data is then stored in a pre-defined profile database for subsequent processing.

[0034] In some embodiments, when the server performs the process of obtaining user demand information and user profile data of the target user corresponding to the user demand information, it may also obtain raw input information, wherein the raw input information includes at least one of the following: user-input consultation text, speech-to-text text, and demand phrase text entered by the salesperson.

[0035] Next, demand-related fields and identity-related fields are extracted from the original input information. The demand-related fields are transformed into user demand information, and the identity-related fields are used to generate the target user's identity information. Then, user profile data is searched in a preset profile database based on the user identity information.

[0036] For example, taking health insurance as an example, User A initiates a dialogue with the intelligent customer service platform via a mobile terminal, inputting the original information: "I spent over 8,000 yuan on hospitalization last year. What kind of insurance should I buy in this situation?" Through speech-to-text transcription, field extraction, and information recognition, the relevant fields for the demand can be determined as: "Core service: Which insurance to buy; Business object: Hospitalization; Time clue: Last year; Amount clue: Over 8,000 yuan." Simultaneously, identity-related fields such as the user ID logged into the intelligent customer service platform are extracted, such as the identity-related field: User A. Then, the demand-related fields are converted into user demand information, and user identity information is generated based on the identity-related fields. Based on this user identity information, user profile data can be queried, such as: Policy information: Holding one "XX Million Medical Insurance" policy, the policy status is valid, the waiting period has passed; Historical claims record: No claims record in the past 2 years; Health declaration: No abnormalities were disclosed at the time of insurance application; Behavioral preferences include preference for self-service processing via the app, having browsed the "Online Claims Guide" article; Tag: Low-risk user, requiring guided service.

[0037] S102. Use the demand decomposition intelligent agent to decompose the core elements from the user demand information, and generate standard demand tags based on the core elements.

[0038] After acquiring user demand information and user profile data, the server can invoke a demand decomposition agent to extract core elements from the user demand information and generate standard demand tags based on these elements. This demand decomposition agent is an AI module configured on the server with autonomous decision-making capabilities and proactive action. The agent can proactively set goals, plan steps, invoke tools, and continuously interact with the environment until the task is completed. Therefore, the demand decomposition agent is a dedicated AI module responsible for decomposing and structuring the ambiguous demands of insurance users. It can transform arbitrary, colloquial, and even incomplete natural language text from user demand information into a set of standardized demand tags that can be directly used by downstream systems.

[0039] like Figure 3 As shown, in some embodiments, when the server executes the requirement decomposition agent to decompose core elements from user requirement information and generate standard requirement tags based on the core elements, it can call the requirement decomposition agent to perform preliminary decomposition of the user requirement information to obtain preliminary decomposition results. Then, based on the preliminary decomposition results, it autonomously plans fuzzy requirement hierarchical extraction subtasks, wherein the requirement decomposition agent is an agent built based on a fine-tuning large model in the insurance field.

[0040] The fine-tuned large-scale model in the insurance field is obtained by secondary training on a general large-scale model using a large amount of data such as insurance dialogues, terms and conditions, and Q&A. To build a demand decomposition intelligent agent, business requirements can be defined first, that is, the core elements, scenario scope, and output format specifications for decomposition are determined. For example, the core elements of decomposition may include the insured population, insurance type, expected coverage amount, budget range, coverage period, and additional value-added service requirements. The scenario scope includes health insurance, life insurance, and auto insurance.

[0041] Next, by collecting real insurance dialogue data and labeling it with demand tags, training and test datasets are constructed. Then, a suitable base model is selected, such as open-source models like Qwen or ChatGLM, or a commercial API. The base model is trained using the training dataset. During training, labeled data can be used to perform LoRA or full-parameter fine-tuning on the base model, enabling it to learn the decomposition logic of the insurance domain. Insurance terms, product manuals, FAQs, and other documents are then sliced ​​and vectorized to construct a vector database, thereby building a RAG knowledge base. When a user asks about a specific product, the system intelligently retrieves relevant information from the knowledge base and then generates an answer based on the retrieval results.

[0042] Based on the defined intelligent agent workflow, tools are designed. The workflow of the requirement decomposition intelligent agent includes task planning, requirement decomposition, and tag output. This allows for the development of API tools for features such as profile querying and product retrieval. After training, the accuracy and recall of the decomposition are evaluated using a test set, and end-to-end integration testing is performed. After testing, the model, knowledge base, and workflow are packaged and deployed to a server to form the requirement decomposition intelligent agent.

[0043] After invoking the demand decomposition intelligent agent, it can autonomously plan sub-tasks for fuzzy demand hierarchical extraction, and decompose core elements from user demand information step by step according to these sub-tasks. These core elements include at least one of the following: insured population, insurance type, expected coverage amount, budget range, coverage period, and additional value-added service requirements. Then, these core elements are converted into standard demand tags, which are structured data tags.

[0044] For example, for a user's demand information that reads, "I want to buy insurance for my elderly family member, preferably with coverage for critical illnesses, with a budget of no more than 5,000 yuan per year, and the coverage amount should be as high as possible. I'll see about other things later," we can call the Agentic intelligent agent to perform information gap analysis. The analysis results for this user demand information are shown in Table 1.

[0045] Table 1. Results of the information gap analysis;

[0046] Based on the information gap analysis results above, the demand decomposition agent can determine that the current information is insufficient to generate complete labels and decides to first decompose the existing information and then actively inquire about the missing items. Therefore, vague demands can be refined layer by layer through hierarchical decomposition. In the first layer of decomposition, the core elements of the insured population can be further refined. For example, "elderly family members" can be decomposed as follows: age range: likely ≥55 years old; health status: unknown (needs further inquiry); relationship with the insured: children insuring their parents, etc. Similarly, in the second layer of decomposition, inferences can be made based on the core elements of the insurance type. For example, based on the dialogue content of "insurance for critical illnesses," the corresponding insurance types in the insurance field can be decomposed into critical illness insurance, medical insurance, cancer insurance, etc. Since critical illness insurance premiums are relatively expensive for the elderly, the combination of "budget of 5,000 yuan / year" and "elderly" can determine that the insurance type is more inclined to be million-yuan medical insurance or cancer medical insurance, etc. By analogy, the above-mentioned layer-by-layer refinement can be carried out for different core elements, thereby transforming vague expressions into standard concepts in the insurance field.

[0047] After the above layer-by-layer decomposition, the demand decomposition intelligent agent can output standardized demand tags, such as standard demand tags in JSON format, like "demand_tags: {Insured population: {Age range: 55-65 years old, Relationship with the insured: Children insuring their parents, Health status: To be supplemented}, Insurance type: {Main category: Health insurance, Subcategory: Million-dollar medical insurance / Cancer prevention medical insurance, Confidence level: Medium-high (needs confirmation)}, Expected sum insured: {Suggested range: 1 million-4 million, Original user statement: As high as possible}, Budget range: {Annual payment limit: 5,000 yuan, Payment method: Annual payment}, Coverage period: {Status: To be supplemented}, Additional value-added services: {Status: To be supplemented, Common options: [Green channel for medical treatment, Hospitalization prepayment, Secondary diagnosis]}}, missing_info: [Coverage period, Additional service requirements, Specific age, Health status], next_action: Follow up on the above missing items}.

[0048] It is evident that the intelligent agent for demand decomposition, built upon a fine-tuned large-scale model in the insurance field, can autonomously plan sub-tasks for hierarchical extraction of fuzzy demands. It can step-by-step extract core elements from natural language, such as the insured population, insurance type, expected coverage amount, budget range, coverage period, and additional value-added service requirements, and output structured standard demand labels. This intelligent agent can transform unstructured user language into structured business data, while also possessing initiative—not passive processing, but proactive planning, hierarchical decomposition, and intelligent completion—providing a precise entry point for the entire intelligent insurance service chain.

[0049] S103. Generate additional requirement tags based on standard requirement tags and user profile data.

[0050] After generating standard requirement tags based on core elements, additional requirement tags can be generated based on the standard requirement tags and user profile data. These additional requirement tags, developed in conjunction with user profile planning, supplement implicit user needs and preferences to refine the requirement analysis results.

[0051] In some embodiments, when the server generates additional demand tags based on standard demand tags and user profile data, it can first create a profile-linked planning subtask and initiate a data call request to the AI ​​customer segmentation system based on the profile-linked planning subtask. This data call request retrieves dynamic tags from the AI ​​customer segmentation system, whereby the dynamic tags include at least one of segmentation / stratification tags, churn risk tags, and insurance intention tags. Then, based on a fine-tuned large-scale model in the insurance field, the standard demand tags and user profile data are cross-analyzed to autonomously plan supplementary inference subtasks.

[0052] Then, historical behavior data is extracted from the user profile data, including historical insurance purchase records and historical claims records. Further reasoning is performed by fusing dynamic tags and historical behavior data through a supplementary reasoning subtask to obtain implicit preferences. Finally, a structured transformation is performed on these implicit preferences to obtain additional demand tags.

[0053] For example, after the demand decomposition AI agent initially extracts explicit demand tags such as target audience, insurance type, and coverage amount, or when a user initiates demand input, the linkage planning AI agent is automatically activated to perform a user profile linkage planning sub-task. During user profile linkage planning, multi-source user profile data can be actively read. Specifically, the linkage planning AI agent proactively initiates data calls to the enterprise's AI user segmentation system to obtain three types of dynamic tags: stratification tags, churn risk tags, and insurance intention tags. Among these, A1-A5 stratification tags can include risk level, value level, and activity level; churn risk tags can include one of high churn probability, medium churn probability, or low churn probability; and insurance intention tags can include one of high purchase inclination, medium purchase inclination, or low purchase inclination.

[0054] Then, historical behavioral data is read and integrated, namely, the user's historical insurance records stored in the unified platform, such as the types of insurance purchased, the sum insured, and the term. Historical claims records are also read, such as the number of claims, the amount of claims, and the type of insurance claimed, to form a baseline of user behavior.

[0055] Then, implicit demand preferences are autonomously planned and reasoned. The linked planning agent, based on a fine-tuned large model, cross-analyzes explicit demand tags with profiles or historical data to autonomously plan and supplement the reasoning logic. For example, if a user is in the A1 stratum (high net worth) and has a high intention to purchase insurance, the autonomous planning will supplement the preference for high-end value-added services, such as international rescue and VIP green channels; if a user has a high number of historical claims, the autonomous planning will suggest increasing the accident insurance coverage amount or adding additional accidental medical liability; if a user has a high risk of churn and has historically only purchased a single type of insurance, the autonomous planning will supplement the preference for multi-type insurance combination discounts to improve retention; if a user does not mention the coverage period, the default period preference will be automatically supplemented based on historical insurance purchase habits, such as if the default period preference has always been 1 year, then the default period preference will be supplemented to 1 year.

[0056] The implicit preferences obtained through inference are then converted into structured supplementary labels, such as {implied_preference: {high_end_service: true,accident_coverage_boost: +50k,combo_discount: true}}. These implicit supplementary labels are then merged with the output explicit requirement labels, deduplicated, and their weights adjusted to form the final complete requirement parsing result set, which is then passed to the downstream Agentic multi-agent product assembly planning unit as input.

[0057] It is evident that by linking user profiles with the planning intelligent agent, the system can proactively read the A1-A5 hierarchical tags, churn risk, and insurance intention tags output by the AI ​​user segmentation system, integrate users' historical insurance and claims data, and autonomously plan and supplement users' implicit needs and preferences to improve the results of demand analysis.

[0058] S104. Use product assembly planning intelligent agent groups to generate product customization results based on standard requirement tags and additional requirement tags.

[0059] After obtaining the standard requirement tags and additional requirement tags, the standard requirement tags and additional requirement tags can be used as input to the product assembly planning intelligent agent group to perform product assembly planning. That is, the product assembly planning intelligent agent group can generate product customization results based on the standard requirement tags and additional requirement tags.

[0060] This includes component planning agents, material planning agents, and solution interpretation agents configured in the server. The component planning agents, material planning agents, and solution interpretation agents can all be constructed by breaking down the agents according to the requirements described in the above embodiments, through training a base model and performing domain fine-tuning. The only difference lies in the training data collected and the corresponding annotations used in constructing the different agents.

[0061] Corresponding to the agents included in the product assembly planning agent group, the product customization results include the product combination output by the component planning agent, the material combination output by the material planning agent, and the interpretation text output by the solution interpretation agent.

[0062] In some embodiments, when the server executes the product assembly planning agent group to generate product customization results based on standard requirement tags and additional requirement tags, it can first use the component planning agent to autonomously plan the execution order of hierarchical matching sub-tasks based on standard requirement tags, execute the hierarchical matching sub-tasks according to the execution order, and obtain product combinations by dynamically adjusting the scheme guarantee strength by integrating additional requirement tags.

[0063] For example, a product assembly planning agent group is an automated product assembly planning unit built using Agentic multi-agent architecture. This group can include three types of collaborative agents: component planning agents, material planning agents, and solution interpretation agents. These agents autonomously complete the entire product assembly process without human intervention in component matching.

[0064] Therefore, the server can pre-build a standardized product component library, that is, a unified standardized component library for all categories of insurance, including travel, health, and home insurance. Components in this standardized insurance component library can include independent coverage liabilities, coverage tiers, value-added services, and supplementary insurance types, and these components can be freely split, combined, and matched. For the component planning intelligent agent, component matching planning can be performed. That is, by receiving structured demand tags output by the parsing engine, it can autonomously plan the hierarchical matching order of main insurance, supplementary insurance, coverage amount, and coverage period, and then dynamically adjust the coverage strength of the plan by integrating customer segmentation profiles, automatically generating product combinations as personalized insurance product plans for target users.

[0065] After the component planning agent outputs the product portfolio, the material planning agent can autonomously plan marketing material generation sub-tasks based on the product portfolio to obtain the material portfolio. For example, if the material planning agent is used for material generation planning, after product assembly is completed, the material planning agent can autonomously plan marketing material generation sub-tasks, scheduling large models to generate layered graphic and text marketing materials and short text introductions suitable for salespersons and customers, thus obtaining the material portfolio. The material planning process can be completed automatically by the agent without manual editing, which can improve the efficiency of material planning.

[0066] After the material planning agent outputs the material combinations, the solution interpretation agent autonomously plans a simplified translation process for complex insurance solutions based on the product and material combinations to obtain interpretive text. For example, the solution interpretation agent can autonomously plan a simplified translation process for complex insurance solutions, breaking down professional insurance terms into easily understandable, conversational interpretive text for customers, which is then simultaneously pushed to agents and customer front-ends.

[0067] It is evident that the multi-dimensional AI customer data intelligent recommendation process based on the fusion of Agentic intelligent agents to perform dynamic programming can increase the conversion rate of business target customer solutions by more than 10%, accurately match customers' differentiated protection needs, and significantly reduce the threshold for creating customized insurance products.

[0068] In some embodiments, after obtaining standard demand tags and additional demand tags, the server can further merge, deduplicate, and dynamically adjust the weights of the standard demand tags and additional demand tags to obtain fused demand information. Then, fused input data is generated based on the fused demand information, wherein the fused input data includes static hierarchical profiles, real-time dialogue semantics, and fused demand information; the static hierarchical profiles are extracted from user profile data; and the real-time dialogue semantics are obtained by performing semantic recognition on the user demand information.

[0069] Then, based on the integrated input data, a multi-dimensional matching degree weighted calculation path is autonomously planned to query products that match the demand in a standardized product component library. This standardized product component library includes standardized product components corresponding to independent coverage responsibilities, coverage tiers, value-added services, and supplementary insurance types. Next, product combinations are extracted from the products that match the demand. Each product combination includes a preset number of insurance products from the products that match the demand.

[0070] For example, if the preset quantity is set to N, the server can be configured with an AI personalized product interpretation and recommendation engine, i.e., a product sales module, which is used to use a recommendation planning agent to perform multi-dimensional intelligent recommendation planning. The recommendation planning agent can integrate three layers of data input, namely customer static hierarchical profiles, real-time dialogue semantic intent, and customer natural language needs, and autonomously plan a multi-dimensional matching degree weighted calculation path to sort and output the TOP N products, forming the optimal personalized customization solution.

[0071] S105. Send the product customization results to the client so that the product customization results can be displayed through the client.

[0072] After generating the product customization results, the server can send the product customization results to the client so that the client can display the product customization results.

[0073] In some embodiments, the insurance product customization system can adopt a unified omnichannel data architecture, which integrates the four major business channels—enterprise online multi-terminal applications, agent service terminals, and internet-based collaborative sales platforms—to uniformly store customer needs, product assembly, and recommendation record data. A single product component library, semantic parsing engine, and Agentic planning intelligent agent synchronously serve all channels, enabling cross-channel interoperability of personalized products for customers.

[0074] The insurance product customization system can also utilize upstream and downstream linkage interfaces. It receives customer tags and dialogue intents from customer segmentation and real-time script systems, and outputs standardized customized product data packages that can be directly underwritten to downstream channel sales systems, achieving full-chain automation from demand analysis and product assembly to sales. By adopting a unified omnichannel data platform architecture and upstream / downstream linkage interfaces, it can unify the product platform, large model analysis engine, and multi-agent planning units to support all online, agent, and cooperative business channels of the enterprise. It allows for the reuse of product components across multiple scenarios such as travel, health, and home, reducing the cost of developing customized systems separately for each channel by 70%.

[0075] By applying the technical solutions of the above embodiments, the insurance product customization method based on natural language intent parsing described in the above embodiments can autonomously and step-by-step parse users' spoken language needs through a demand decomposition intelligent agent, and automatically assemble products through a component planning intelligent agent, and generate material combinations through a matching material planning intelligent agent, thereby improving the richness and accuracy of insurance product-related information.

[0076] In some embodiments, as a refinement and extension of the specific implementation of the above embodiments, and to fully illustrate the specific implementation process of this embodiment, some embodiments of this application also provide a method for customizing insurance products based on natural language intent parsing. The difference between this method and the above embodiments is that it can support secondary custom adjustments to the coverage. For example... Figure 4 As shown, the method includes: S201, Obtain service adjustment instructions; S202. Parse adjustment items from business adjustment instructions; S203. Modify the product customization results according to the adjustment items.

[0077] After the product assembly planning intelligent agent group generates product customization results based on standard and additional requirement tags, the server can also obtain business adjustment instructions. These business adjustment instructions are control instructions sent by the business client, i.e., manual adjustment instructions input by business personnel through the business client regarding the product customization results.

[0078] After receiving the business adjustment instruction, the server can parse the adjustment items from the business adjustment instruction. The adjustment items include product screening rules for product portfolios, material constraint rules for material portfolios, and guarantee responsibility information for the interpretation text.

[0079] Then, modify the product customization results according to the adjustment items. Specifically, for the product combination in the product customization results, the insurance products in the product combination can be replaced or added / deleted according to the product screening rules in the adjustment items; for the material combination in the product customization results, the material content can be modified according to the material constraint rules in the adjustment items; and for the interpretation text in the product customization results, the information can be supplemented according to the coverage liability information in the adjustment items, so that the salesperson can manually customize and adjust the coverage liability a second time.

[0080] By applying the technical solutions of the above embodiments, the insurance product customization method based on natural language intent parsing described in the above embodiments can flexibly modify the product customization results through customization adjustments, thereby adapting to multiple business scenarios. For example, the large model and Agentic planning agent can be migrated to all non-motor insurance categories such as medical insurance, home insurance, and travel insurance, supporting DIY customization for multiple roles such as individual customers, corporate group orders, and sales teams, with strong technology reusability.

[0081] Furthermore, it can optimize the customer experience by enabling the intelligent agent to autonomously break down complex insurance terms and provide customized adjustment instructions, thus lowering the barrier to understanding for customers. Moreover, by autonomously planning personalized product solutions and reducing the push of homogenized packages, it helps increase customers' willingness to purchase insurance.

[0082] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0083] In some embodiments, as a specific implementation of the insurance product customization method based on natural language intent parsing described in the above embodiments, some embodiments of this application also provide an insurance product customization system based on natural language intent parsing, such as... Figure 5 As shown, the system includes: The client is configured to perform user interactions, receiving user input and displaying product customization results; The server establishes a communication connection with the client. The server is configured with a group of intelligent agents for demand breakdown and product assembly planning. The product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent. The server is configured as follows: Obtain user demand information and user profile data of the target users corresponding to the user demand information; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The standard requirement tags and the additional requirement tags are input into the product assembly planning intelligent agent group to obtain the product customization results output by the product assembly planning intelligent agent group. The product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent. The product customization results are sent to the client so that the product customization results can be displayed through the client.

[0084] In some embodiments, the server is further configured to acquire user demand information and user profile data of the target user corresponding to the user demand information, and to perform the following operations: Obtain raw input information, which includes at least one of the following: user-input consultation text, speech-to-text text, and short requirement sentences entered by salesperson; Extract the requirement-related fields and identity-related fields from the original input information; The requirement association field is transformed into the user requirement information, and the user identity information of the target user is generated based on the identity association field; The user profile data is retrieved from a preset profile database based on the user's identity information.

[0085] In some embodiments, the server, which executes a demand decomposition agent to decompose core elements from the user demand information and generate standard demand tags based on the core elements, is further configured to: The requirement decomposition intelligent agent is invoked to perform a preliminary decomposition of the user requirement information to obtain preliminary decomposition results; Based on the preliminary decomposition results, the system autonomously plans fuzzy demand hierarchical extraction sub-tasks, and the demand decomposition intelligent agent is an intelligent agent built based on a fine-tuned large model in the insurance field; According to the fuzzy requirements, the sub-tasks are extracted step by step to decompose the core elements from the user requirements information. The core elements include at least one of the following: insured population, type of insurance, expected coverage amount, budget range, coverage period, and additional value-added service requirements. The core elements are transformed into standard requirement tags, which are structured data tags.

[0086] In some embodiments, the server is further configured to generate additional demand tags based on the standard demand tags and the user profile data. Create a profile-linked planning sub-task; Based on the profile linkage planning sub-task, a data call request is initiated to the AI ​​customer segmentation system to obtain dynamic tags. The dynamic tags include at least one of segmentation and stratification tags, churn risk tags, and insurance intention tags. Based on the fine-tuning of the large model in the insurance field, the standard demand tags and the user profile data are cross-analyzed to autonomously plan supplementary reasoning sub-tasks; Historical behavior data is extracted from the user profile data, including historical insurance purchase records and historical claims records; The implicit preferences are obtained by fusing the dynamic tags and historical behavior data through the supplementary reasoning subtask. Perform a structured transformation on the implicit preferences to obtain the additional demand labels.

[0087] In some embodiments, the server, which executes the generation of product customization results using a group of product assembly planning agents based on the standard requirement tags and the additional requirement tags, is further configured to: The component is used to plan the execution order of hierarchical subtasks autonomously based on the standard requirement tags; The product portfolio is obtained by executing the layered sub-tasks in the order described, and by dynamically adjusting the guarantee strength of the scheme by integrating the additional demand tags. The material planning agent autonomously plans marketing material generation sub-tasks based on the product portfolio to obtain the material portfolio. The solution uses an intelligent agent to autonomously plan a simplified translation process for complex insurance schemes based on the product portfolio and the material portfolio, in order to obtain the interpreted text.

[0088] In some embodiments, the server is further configured to: The standard demand tags and the additional demand tags are merged, deduplicated, and dynamically weighted to obtain fused demand information; Fusion input data is generated based on the fusion requirement information. The fusion input data includes static hierarchical profiles, real-time dialogue semantics, and the fusion requirement information. The static hierarchical profiles are extracted from the user profile data. The real-time dialogue semantics are obtained by performing semantic recognition on the user requirement information. Based on the fused input data, a multi-dimensional matching degree weighted calculation path is autonomously planned to query products that match the needs in the standardized product component library. The standardized product component library includes standardized product components corresponding to independent protection liability, coverage amount level, value-added services, and supplementary insurance types. Extract the product portfolio from the products that meet the demand, the product portfolio including a preset number of insurance products from the products that meet the demand.

[0089] In some embodiments, the server is further configured to: Obtain a service adjustment instruction, wherein the service adjustment instruction is a control instruction sent by the service client; The adjustment items are parsed from the business adjustment instruction. The adjustment items include product screening rules for the product portfolio, material constraint rules for the material portfolio, and guarantee responsibility information for the interpretation text. Modify the product customization results according to the adjustment items described.

[0090] By applying the technical solutions of the above embodiments, in the insurance product customization system based on natural language intent parsing described above, the server can obtain user demand information and target user profile data through the client, and use a demand decomposition intelligent agent to decompose core elements from the user demand information, and generate standard demand tags based on the core elements. Then, additional demand tags are generated based on the standard demand tags and user profile data. Finally, a product assembly planning intelligent agent group generates product customization results based on the standard demand tags and additional demand tags, and sends the product customization results to the client for display. The system can autonomously and step-by-step parse users' spoken demands through the demand decomposition intelligent agent, automatically assemble products through the component planning intelligent agent, and generate material combinations through the matching material planning intelligent agent, thereby improving the richness and accuracy of insurance product-related information.

[0091] In some embodiments, a computer device is provided, which may be a server. The computer device includes a processor, memory, a network interface, and a database connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of the method on the server side.

[0092] In some embodiments, a computer device is provided, which can be a client. The computer device includes a processor, memory, a network interface, a display screen, and an input device connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of the method on the client side.

[0093] In some embodiments, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Obtain user demand information and user profile data of the target user corresponding to the user demand information, wherein the user demand information includes natural language text input from multiple sources; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The product assembly planning intelligent agent group generates product customization results based on the standard requirement tags and the additional requirement tags; the product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent; the product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent; The product customization results are sent to the client so that the product customization results can be displayed through the client.

[0094] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Obtain user demand information and user profile data of the target user corresponding to the user demand information, wherein the user demand information includes natural language text input from multiple sources; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The product assembly planning intelligent agent group generates product customization results based on the standard requirement tags and the additional requirement tags; the product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent; the product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent; The product customization results are sent to the client so that the product customization results can be displayed through the client.

[0095] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0096] 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 a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0097] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0098] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0099] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. An insurance product customization system based on natural language intent parsing, characterized in that, The system includes: The client is configured to perform user interactions, receiving user input and displaying product customization results; The server establishes a communication connection with the client. The server is configured with a group of intelligent agents for demand breakdown and product assembly planning. The product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent. The server is configured as follows: Obtain user demand information and user profile data of the target users corresponding to the user demand information; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The standard requirement tags and the additional requirement tags are input into the product assembly planning intelligent agent group to obtain the product customization results output by the product assembly planning intelligent agent group. The product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent. The product customization results are sent to the client so that the product customization results can be displayed through the client.

2. The system according to claim 1, characterized in that, The server is configured to retrieve user demand information and user profile data of the target user corresponding to the user demand information, and is further configured to: Obtain raw input information, which includes at least one of the following: user-input consultation text, speech-to-text text, and short requirement sentences entered by salesperson; Extract the requirement-related fields and identity-related fields from the original input information; The requirement association field is transformed into the user requirement information, and the user identity information of the target user is generated based on the identity association field; The user profile data is retrieved from a preset profile database based on the user's identity information.

3. The system according to claim 1, characterized in that, The server, which executes a user requirement decomposition agent to extract core elements from the user requirement information and generate standard requirement tags based on the core elements, is also configured to: The requirement decomposition intelligent agent is invoked to perform a preliminary decomposition of the user requirement information to obtain preliminary decomposition results; Based on the preliminary decomposition results, the system autonomously plans fuzzy demand hierarchical extraction sub-tasks, and the demand decomposition intelligent agent is an intelligent agent built based on a fine-tuned large model in the insurance field; According to the fuzzy requirements, the sub-tasks are extracted step by step to decompose the core elements from the user requirements information. The core elements include at least one of the following: insured population, type of insurance, expected coverage amount, budget range, coverage period, and additional value-added service requirements. The core elements are transformed into standard requirement tags, which are structured data tags.

4. The system according to claim 1, characterized in that, The server is configured to generate additional demand tags based on the standard demand tags and the user profile data, and is also configured to: Create a profile-linked planning sub-task; Based on the profile linkage planning sub-task, a data call request is initiated to the AI ​​customer segmentation system to obtain dynamic tags. The dynamic tags include at least one of segmentation and stratification tags, churn risk tags, and insurance intention tags. Based on the fine-tuning of the large model in the insurance field, the standard demand tags and the user profile data are cross-analyzed to autonomously plan supplementary reasoning sub-tasks; Historical behavior data is extracted from the user profile data, including historical insurance purchase records and historical claims records; The implicit preferences are obtained by fusing the dynamic tags and historical behavior data through the supplementary reasoning subtask. Perform a structured transformation on the implicit preferences to obtain the additional demand labels.

5. The system according to claim 1, characterized in that, The server executes the product assembly planning intelligent agent group to generate product customization results based on the standard requirement tags and the additional requirement tags, and is also configured to: The component is used to plan the execution order of hierarchical subtasks autonomously based on the standard requirement tags; The product portfolio is obtained by executing the layered sub-tasks in the order described, and by dynamically adjusting the guarantee strength of the scheme by integrating the additional demand tags. The material planning agent autonomously plans marketing material generation sub-tasks based on the product portfolio to obtain the material portfolio. The solution uses an intelligent agent to autonomously plan a simplified translation process for complex insurance schemes based on the product portfolio and the material portfolio, in order to obtain the interpreted text.

6. The system according to claim 5, characterized in that, The server is also configured to: The standard demand tags and the additional demand tags are merged, deduplicated, and dynamically weighted to obtain fused demand information; Based on the fusion requirement information, fusion input data is generated, which includes static hierarchical profiles, real-time dialogue semantics, and the fusion requirement information. The static layered profile is extracted from the user profile data; The real-time dialogue semantics are obtained by performing semantic recognition on the user's demand information; Based on the fused input data, a multi-dimensional matching degree weighted calculation path is autonomously planned to query products that match the needs in the standardized product component library. The standardized product component library includes standardized product components corresponding to independent protection liability, coverage amount level, value-added services, and supplementary insurance types. Extract the product portfolio from the products that meet the demand, the product portfolio including a preset number of insurance products from the products that meet the demand.

7. The system according to claim 1, characterized in that, The server is also configured to: Obtain a service adjustment instruction, wherein the service adjustment instruction is a control instruction sent by the service client; The adjustment items are parsed from the business adjustment instruction. The adjustment items include product screening rules for the product portfolio, material constraint rules for the material portfolio, and guarantee responsibility information for the interpretation text. Modify the product customization results according to the adjustment items described.

8. A method for customizing insurance products based on natural language intent parsing, characterized in that, Applied to the system according to any one of claims 1-7, the method comprises: Obtain user demand information and user profile data of the target users corresponding to the user demand information; The demand decomposition agent is used to decompose the core elements from the user demand information and generate standard demand tags based on the core elements. Additional demand tags are generated based on the standard demand tags and the user profile data; The standard requirement tags and the additional requirement tags are input into the product assembly planning intelligent agent group to obtain the product customization results output by the product assembly planning intelligent agent group; the product assembly planning intelligent agent group includes a component planning intelligent agent, a material planning intelligent agent, and a solution interpretation intelligent agent; the product customization results include the product combination output by the component planning intelligent agent, the material combination output by the material planning intelligent agent, and the interpretation text output by the solution interpretation intelligent agent; The product customization results are sent to the client so that the product customization results can be displayed through the client.

9. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, The processor implements the method of claim 8 when executing the computer program.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of claim 8.

Images