An intelligent insurance service system based on a vertical field model and a large language model
By combining vertical domain models and large language models, the intelligent insurance service system solves the problems of high labor costs, low efficiency, and insufficient accuracy of professional Q&A in the traditional insurance service model. It realizes full-process automation and integration of insurance and health management, supports cross-scenario business processing, and improves user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING ZHONGKE WANGUO INTERNET TECH CO LTD
- Filing Date
- 2026-05-14
- Publication Date
- 2026-07-07
AI Technical Summary
Traditional insurance service models suffer from high labor costs, low service efficiency, insufficient accuracy of professional Q&A, cumbersome underwriting and claims processes, disconnect between insurance and health services, difficulty in maintaining knowledge bases, and fragmented service scenarios. Furthermore, general-purpose language models lack in-depth training in the insurance and medical vertical fields, making them unable to handle complex business needs across different scenarios.
An intelligent insurance service system based on vertical domain models and large language models is adopted. By combining general large language models with insurance and health vertical domain models, a multi-agent collaborative mechanism is constructed to realize full-process automated services, including intelligent operation, marketing and health management modules. It supports multi-terminal access and completes complex business tasks across scenarios through multi-agent collaboration.
It has achieved intelligent and integrated insurance services throughout the entire process, reduced labor costs, improved service efficiency and the accuracy of professional Q&A, integrated insurance and health management, supported complex business processing across scenarios, provided services throughout the entire life cycle, and improved user satisfaction.
Smart Images
Figure CN122347477A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of artificial intelligence technology, specifically to an intelligent insurance service system based on a vertical domain model and a large language model. Background Technology
[0002] As the digital transformation of the insurance industry deepens, customers' demands for the timeliness, professionalism, and personalization of insurance services continue to rise. Traditional insurance service models suffer from several insurmountable pain points: traditional human customer service requires 24 / 7 support, with labor costs exceeding 30% of insurance operating costs; insurance business is highly specialized, requiring 3-6 months of training for human customer service representatives, and the varying levels of expertise among different personnel make it difficult to standardize service quality; traditional underwriting and claims processing rely on manual review of large amounts of paper or electronic documents, with an average claims processing cycle of 3-7 working days, and even up to one month for complex cases; furthermore, inconsistent manual review standards easily lead to errors and omissions in claims. Although some insurance customer service systems based on general large language models have emerged in recent years, they generally suffer from the following shortcomings: they only use general large models, lacking in-depth training in the insurance and medical vertical fields, resulting in a professional question-and-answer accuracy rate of less than 90%, and failing to achieve full automation of core business processes such as underwriting, claims processing, and health management; the knowledge base cannot be automatically updated, resulting in high operating costs; and they lack multi-agent collaboration mechanisms, making it impossible to handle complex business needs across different scenarios.
[0003] Therefore, this invention proposes an intelligent insurance service system that integrates a general-purpose large language model and a vertical domain model to achieve intelligent and integrated insurance services throughout the entire process. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides an intelligent insurance service system based on a vertical domain model and a large language model. This system solves problems such as high labor costs, low service efficiency, insufficient accuracy of professional Q&A, cumbersome underwriting and claims processes, disconnect between insurance and health services, difficulty in maintaining the knowledge base, and fragmented service scenarios in existing insurance service systems.
[0005] To achieve the above objectives, the present invention provides the following technical solution: An intelligent insurance service system based on a vertical domain model and a large language model includes a unified access layer, an application layer, a capability layer, a base layer, and a service monitoring and maintenance module. The base layer includes a general large language model, an insurance vertical domain model, a health vertical domain model, and a multi-source data storage module, which are used to provide basic semantic understanding, content generation, and data storage capabilities. The capability layer includes a multi-agent collaboration module, an intelligent knowledge base module, a context storage module, a capability management module, and a basic AI service module, which are used to build the core functional components required for insurance services based on the capabilities of the base layer; The application layer includes an intelligent operation module, an intelligent marketing module, and an intelligent health management module, which are used to cover all business scenarios of insurance services and provide customers with integrated insurance and health management services. The unified access layer is used to support unified traffic access for multiple terminals, thereby achieving full terminal service coverage. The service monitoring and maintenance module is used to provide 24-hour system monitoring and maintenance services and supports a four-level fault response mechanism.
[0006] Furthermore, the insurance vertical domain model of the base layer is based on a general large language model, which has general semantic understanding, content generation, logical reasoning and multi-turn dialogue capabilities. It is suitable for question-and-answer and dialogue scenarios in general domains. It is obtained by fine-tuning instructions and aligning with human feedback through an insurance industry-specific corpus. The health vertical domain model of the base layer is based on a general large language model and is trained with massive authoritative medical data. It has the capabilities of health consultation, initial analysis of disease conditions, triage and diagnosis, and chronic disease management.
[0007] Furthermore, the multi-agent collaboration module of the capability layer includes a customer service agent, an underwriting agent, a claims agent, a health management agent, and a marketing agent. Each agent calls upon basic AI services and base layer model capabilities through the capability management module to collaboratively complete complex business tasks across scenarios.
[0008] Furthermore, the specific workflow steps of the intelligent knowledge base module in the capability layer are as follows: Step 1: Knowledge Acquisition and Compilation; It supports importing various file formats and web page content, and automatically extracts knowledge from the original materials using large model technology to generate individual knowledge points; Step 2: Knowledge Extraction; The original text is processed using a large model to extract key information and form a structured knowledge representation, which supports manual annotation. Step 3: Automatically generate frequently asked questions and answer pairs; Based on the extracted knowledge, frequently asked questions and answer pairs are automatically generated, and similar question sentences are generated using a large model; Step 4: Knowledge Verification; It supports manual review and correction of automatically generated FAQs and answers, and publishes them to the knowledge base after approval.
[0009] Furthermore, the basic AI service modules of the capability layer include OCR image recognition service, ASR speech recognition service, TTS speech synthesis service, entity information recognition service, user emotion recognition service, text classification service, sentence similarity calculation service, and sensitive information recognition service.
[0010] Furthermore, the intelligent operation module of the application layer is used to automate the entire policy service process, including dialogue communication, customer policy inquiry, policy maintenance, claims processing, renewal and new policy processing, and intelligent quality inspection functions.
[0011] Furthermore, the intelligent marketing module in the application layer is used to collect customer information through multi-round dialogues, build customer profiles, conduct risk assessments and protection gap analysis, and provide customers with personalized insurance product recommendations, comparative analysis, and intelligent underwriting services.
[0012] Furthermore, the intelligent health management module in the application layer is used to provide customers with health consultation, health assessment, initial analysis of symptoms, triage and guidance, health management service recommendations, online appointments, and chronic disease care services.
[0013] This invention provides an intelligent insurance service system based on a vertical domain model and a large language model. It has the following beneficial effects: 1. This invention provides an intelligent insurance service system based on a vertical domain model and a large language model. It adopts an architecture that combines a general large language model with insurance and health vertical domain models. The general large model is responsible for general semantic understanding and content generation, while the vertical domain model is responsible for accurate answers based on professional knowledge. It also provides fully automated services, realizing full automation of core businesses such as policy inquiry, policy maintenance, claims, renewal, and underwriting, which significantly reduces labor costs.
[0014] 2. This invention provides an intelligent insurance service system based on a vertical domain model and a large language model. Through the collaborative work of multiple professional intelligent agents, it can handle complex business tasks across scenarios, deeply integrate insurance services with health management, and provide customers with full life-cycle services from health consultation and disease guidance to insurance coverage and chronic disease management. It supports access from multiple terminals and covers core scenarios such as product marketing, policy services, customer service, policy maintenance services, claims services, and health management. The service process is seamless, thereby significantly improving user satisfaction. Attached Figure Description
[0015] Figure 1 This is a system functional block diagram of the intelligent insurance service system based on a vertical domain model and a large language model according to the present invention; Figure 2 This is a flowchart showing the detailed steps of the intelligent insurance service system based on a vertical domain model and a large language model according to the present invention. Detailed Implementation
[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0017] Example 1: like Figure 1 As shown, this embodiment of the invention provides an intelligent insurance service system based on a vertical domain model and a large language model, including a unified access layer, an application layer, a capability layer, a base layer, and a service monitoring and maintenance module. The base layer includes a general large language model, an insurance vertical domain model, a health vertical domain model, and a multi-source data storage module, which are used to provide basic semantic understanding, content generation, and data storage capabilities. The capability layer includes a multi-agent collaboration module, an intelligent knowledge base module, a context storage module, a capability management module, and a basic AI service module, which are used to build the core functional components required for insurance services based on the capabilities of the base layer; The application layer includes an intelligent operation module, an intelligent marketing module, and an intelligent health management module, which are used to cover all business scenarios of insurance services and provide customers with integrated insurance and health management services. The unified access layer is used to support unified traffic access for multiple terminals, thereby achieving full terminal service coverage. The service monitoring and maintenance module is used to provide 24-hour system monitoring and maintenance services and supports a four-level fault response mechanism. The base layer provides the system with basic modeling and data storage capabilities, including: General Large Language Model: Employs a pre-trained large-scale language model with hundreds of billions of parameters, possessing general semantic understanding, content generation, logical reasoning, and multi-turn dialogue capabilities, suitable for question answering and dialogue scenarios in general domains; Insurance Vertical Domain Model: Based on a general large language model, it is trained by fine-tuning instructions and human feedback alignment (RLHF) through an insurance industry-specific corpus (including insurance terms, underwriting and claims rules, policy business processes, insurance knowledge popularization, historical claims cases, etc.). The accuracy rate is ≥98% in tasks such as insurance professional knowledge Q&A and underwriting and claims decision-making. Health vertical domain model: Based on a general large language model, it is trained by massive authoritative medical data (including disease knowledge of various departments, medical records, treatment plans, expert experience, drug instructions, etc.), with a medical knowledge coverage rate of ≥98% and a medical knowledge accuracy rate of ≥90%. It has the ability to provide health consultation, initial analysis of disease, triage and patient guidance and chronic disease management. Multi-source data storage module: Supports multiple relational databases such as PostgreSQL, Oracle, SQL Server, and MySQL, as well as unstructured data formats such as JSON and HTML, for storing policy information, work order information, customer information, medical knowledge, insurance knowledge, session data, and model training data.
[0018] The capability layer builds upon the capabilities of the base layer to construct the core functional components required for insurance services, including: Multi-agent collaboration module: including customer service agent, underwriting agent, claims agent, health management agent and marketing agent. Each agent has independent task processing capabilities. Through the capability management module, they call basic AI services and base layer model capabilities to collaboratively complete complex business tasks across scenarios. Intelligent Knowledge Base Module: Used to build, manage, and iterate knowledge bases in the insurance and health fields, supporting import of various file formats, automatic knowledge extraction, automatic FAQ generation, automatic generation of similar questions, and generative question answering; Context storage module: It uses distributed caching technology to store context information of customer-system interaction, supports semantic understanding and intent recognition for up to 50 rounds of dialogue, and realizes a coherent multi-turn dialogue experience; Capability Management Module: Used to manage various capability interfaces of the system in a unified manner, including basic AI service interfaces, model call interfaces and third-party business system interfaces, to achieve unified scheduling of capabilities, access management and traffic control; Basic AI service modules include OCR image recognition service, ASR speech recognition service, TTS speech synthesis service, entity information recognition service, user emotion recognition service, text classification service, sentence similarity calculation service, and sensitive information recognition service.
[0019] The workflow of the intelligent knowledge base module is as follows: Step 1: Knowledge Acquisition and Compilation Supports batch import of various file formats such as Word, Excel, PDF, PPT, TXT, MP3, MP4, PNG, Visio, and web page content. It automatically extracts knowledge from the original materials using large model technology, generates single knowledge points, and fully preserves the images, text, and layout format of the materials. Step 2: Knowledge Extraction This approach leverages a large-scale model to perform word segmentation, part-of-speech tagging, and named entity recognition on the original text. It extracts information such as question pairs, context, and key parameters to form a structured knowledge representation. Simultaneously, it supports manual annotation of knowledge block scope, answer priority, and keywords, improving the accuracy of knowledge extraction. Step 3: FAQ generation: Based on extracted and labeled knowledge, it automatically generates common questions and answers (FAQs) and uses a large model to generate similar question sentences, supporting the addition of expanded words and custom question wording. Step 4: Knowledge Verification It provides a visual review interface, supporting manual review, correction and supplementation of generated FAQs. Once approved, it is automatically published to the knowledge base and a publication update log is generated. Step 5: Knowledge Iteration By analyzing historical dialogue data between customers and the chatbot, user likes / dislikes feedback, and trending questions, the system uses a large model for automatic clustering and text mining to identify unknown and high-frequency questions, and automatically updates the knowledge base. At the same time, the model is regularly fine-tuned using new business data to continuously improve the accuracy of question answering.
[0020] The application layer covers all business scenarios of insurance services, providing customers with integrated insurance and health management services, including: Intelligent Operation Module: Enables full automation of policy services, including dialogue communication, customer policy inquiry, policy maintenance, claims processing, renewal / new policy processing, and intelligent quality inspection. It supports a service mode of one-click online consultation and jump-to-processing, allowing customers to complete more than 90% of routine policy business without human intervention. Intelligent Marketing Module: Collects basic customer information, health status, financial status, and insurance needs through multi-round dialogues to build customer profiles. Based on these profiles, it conducts health risk assessments, financial risk assessments, and protection gap analysis, recommends suitable insurance products to customers, and provides comparative analysis. It also supports intelligent pre-underwriting, automatically generates underwriting conclusions, and notifies customers via SMS, email, or APP. The intelligent health management module provides customers with health consultation, health assessment, initial disease analysis, triage and guidance, health management service recommendations, online appointment and chronic disease care services. It simulates the doctor's consultation process, collects patient disease information through multiple rounds of dialogue, automatically generates medical guidance and electronic medical records, and supports integration with hospital systems to provide green channel registration services.
[0021] The unified access layer supports unified traffic access from various terminals such as PC official website, APP, WeChat Work, WeChat Official Account, WeChat Mini Program, H5 and telephone, achieving full terminal service coverage, and customers can obtain a consistent service experience through any terminal.
[0022] The service monitoring and maintenance module provides 24 / 7 system monitoring and maintenance services, including system performance monitoring (CPU, memory, disk, network), business data monitoring (PV, UV, order volume, conversion rate), security monitoring (sensitive information leakage, malicious attacks), and fault early warning and handling functions; it supports a four-level fault response mechanism, including: Level 1 Emergency Fault (Basic System Functions Cannot Operate): Response within 5 minutes; if the problem cannot be resolved within 1 hour, on-site support will be provided. Level 2 critical fault (critical function abnormality): Response within 10 minutes; if the problem cannot be resolved within 1 hour, a technical expert will be dispatched to handle it on-site. Level 3 General Issues (Usability Issues): Response within 15 minutes; if not resolved within 2 hours, escalation to technical experts. Level 4 General Requirement (Function Optimization Suggestion): Provide a solution within 2 days.
[0023] Example 2: like Figure 2 As shown, this embodiment of the invention provides an intelligent insurance service system based on a vertical domain model and a large language model. The specific steps of the system's health insurance service process are as follows: Step 1: Customers access the smart insurance butler system via WeChat mini-program and enter their questions; Step 2: The system's unified access layer receives customer requests and forwards them to the intelligent marketing module and intelligent health management module in the application layer; Step 3: The intelligent health management module calls the health management intelligent agent to collect information such as the customer's insomnia symptoms, duration, accompanying symptoms, medical history and lifestyle habits through multiple rounds of dialogue, and generates a preliminary health assessment report; Step 4: The intelligent marketing module calls the underwriting intelligent agent to automatically generate an underwriting conclusion based on the customer's health assessment report and medical insurance underwriting rules; Step 5: Based on customer profiles and underwriting conclusions, the intelligent marketing module recommends suitable medical insurance products to customers and provides product comparison analysis; Step 6: The customer selects one of the products, and the system automatically guides the customer to complete the online insurance application and payment; Step 7: After successful insurance application, the intelligent health management module provides customers with health advice and sleep monitoring services to improve insomnia, and regularly tracks the customer's health status.
[0024] The following points should be noted in this article: 1. The accompanying drawings of the embodiments disclosed herein only relate to the structures involved in the embodiments disclosed herein; other structures can be referred to in general design.
[0025] 2. Where there is no conflict, the embodiments of this disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.
[0026] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
Claims
1. An intelligent insurance service system based on a vertical domain model and a large language model, comprising a unified access layer, an application layer, a capability layer, a foundation layer, and a service monitoring and maintenance module, characterized in that: The base layer includes a general-purpose large language model, an insurance vertical domain model, a health vertical domain model, and a multi-source data storage module, which are used to provide basic semantic understanding, content generation, and data storage capabilities. The capability layer includes a multi-agent collaboration module, an intelligent knowledge base module, a context storage module, a capability management module, and a basic AI service module, which are used to build the core functional components required for insurance services based on the capabilities of the base layer; The application layer includes an intelligent operation module, an intelligent marketing module, and an intelligent health management module, which are used to cover all business scenarios of insurance services and provide customers with integrated insurance and health management services. The unified access layer is used to support unified traffic access for multiple terminals, thereby achieving full terminal service coverage. The service monitoring and maintenance module is used to provide 24-hour system monitoring and maintenance services and supports a four-level fault response mechanism.
2. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The insurance vertical domain model of the base layer is based on a general large language model, which has the ability of general semantic understanding, content generation, logical reasoning and multi-turn dialogue. It is suitable for question-and-answer and dialogue scenarios in general domains. It is obtained by fine-tuning instructions and aligning with human feedback through a corpus dedicated to the insurance industry. The health vertical domain model of the base layer is based on a general large language model and is trained with massive authoritative medical data. It has the ability of health consultation, initial analysis of disease, triage and diagnosis and chronic disease management.
3. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The multi-agent collaboration module of the capability layer includes a customer service agent, an underwriting agent, a claims agent, a health management agent, and a marketing agent. Each agent calls basic AI services and base layer model capabilities through the capability management module to collaboratively complete complex business tasks across scenarios.
4. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The specific workflow steps of the intelligent knowledge base module in the capability layer are as follows: Step 1: Knowledge Acquisition and Compilation; It supports importing various file formats and web page content, and automatically extracts knowledge from the original materials using large model technology to generate individual knowledge points; Step 2: Knowledge Extraction; The original text is processed using a large model to extract key information and form a structured knowledge representation, which supports manual annotation. Step 3: Automatically generate frequently asked questions and answer pairs; Based on the extracted knowledge, frequently asked questions and answer pairs are automatically generated, and similar question sentences are generated using a large model; Step 4: Knowledge Verification; It supports manual review and correction of automatically generated FAQs and answers, and publishes them to the knowledge base after approval.
5. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The basic AI service modules of the capability layer include OCR image recognition service, ASR speech recognition service, TTS speech synthesis service, entity information recognition service, user emotion recognition service, text classification service, sentence similarity calculation service, and sensitive information recognition service.
6. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The intelligent operation module of the application layer is used to automate the entire policy service process, including dialogue communication, customer policy inquiry, policy maintenance, claims processing, renewal and new policy processing, and intelligent quality inspection.
7. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The intelligent marketing module in the application layer is used to collect customer information through multi-round dialogues, build customer profiles, conduct risk assessments and protection gap analysis, and provide customers with personalized insurance product recommendations, comparative analysis and intelligent underwriting services.
8. The intelligent insurance service system based on a vertical domain model and a large language model according to claim 1, characterized in that: The intelligent health management module in the application layer is used to provide customers with health consultation, health assessment, initial analysis of symptoms, triage and guidance, health management service recommendations, online appointments, and chronic disease care services.