Underwriting suggestion generation method and device, equipment, medium and program product

Abstract

The invention relates to the technical field of computers, in particular to an underwriting suggestion generation method and device, equipment, a medium and a program product. The method comprises the following steps: acquiring underwriting inquiry information, and analyzing the underwriting inquiry information to obtain demand information corresponding to the underwriting inquiry information; an underwriting knowledge base is searched for underwriting information corresponding to the demand information, underwriting key point prompt information is generated according to the underwriting information, and the underwriting knowledge base comprises retrieval enhancement generation underwriting information; and according to the retrieval enhancement generation underwriting information and the underwriting key point prompt information, generating an underwriting suggestion. By adopting the scheme, personalized underwriting suggestions can be generated, specific requirements and risk conditions of different insurance applicants can be met, and the underwriting efficiency and accuracy are relatively high.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a method, device, equipment, medium and program product for generating underwriting recommendations. Background Art

[0002] In the insurance industry, underwriting is an important part of the underwriting process. In the existing underwriting process, underwriters need to conduct a comprehensive assessment based on the insured's personal information, health status, occupational risks and other factors. This method is highly dependent on manual labor and requires underwriters to manually review a large amount of data, which is not only time-consuming and labor-intensive, but also inefficient, and prone to misjudgment or inconsistent standards due to human factors. In addition, existing technologies lack personalized underwriting recommendations and cannot meet the specific needs and risk conditions of different insured persons. Summary of the invention

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

[0004] To this end, the first purpose of the present disclosure is to propose a method for generating underwriting recommendations to generate personalized underwriting recommendations that meet the specific needs and risk conditions of different policyholders, and have high underwriting efficiency and accuracy.

[0005] The second objective of the present disclosure is to provide an underwriting recommendation generating device.

[0006] A third objective of the present disclosure is to provide an electronic device.

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

[0008] A fifth object of the present disclosure is to provide a computer program product.

[0009] To achieve the above-mentioned purpose, the first embodiment of the present disclosure proposes a method for generating an underwriting recommendation, comprising:

[0010] Acquire underwriting inquiry information, parse the underwriting inquiry information, and obtain demand information corresponding to the underwriting inquiry information;

[0011] Searching for underwriting information corresponding to the demand information in an underwriting knowledge base, and generating underwriting key point prompt information according to the underwriting information, wherein the underwriting knowledge base includes retrieval enhancement to generate underwriting information;

[0012] The underwriting information and the underwriting key point prompt information are generated based on the search enhancement, and an underwriting recommendation is generated.

[0013] Optionally, before searching the underwriting information corresponding to the demand information in the underwriting knowledge base, the method further includes:

[0014] Acquire search enhancement to generate underwriting information, and collect policyholder related information from at least one data source;

[0015] Performing data integration on the underwriting information generated by the search enhancement and the policyholder related information to obtain integrated information;

[0016] The integrated information is subjected to data preprocessing to obtain preprocessed integrated information, and an underwriting knowledge base is constructed based on the preprocessed integrated information.

[0017] Optionally, searching an underwriting knowledge base for underwriting information corresponding to the demand information includes:

[0018] Recalling the search information corresponding to the demand information from the underwriting knowledge base using the search enhancement generation technology;

[0019] Performing intent recognition on the search information, and performing service enhancement on the search information according to the intent recognition result to obtain service enhancement information;

[0020] Generate underwriting information corresponding to the demand information based on the search information and the service enhancement information.

[0021] Optionally, generating underwriting key point prompt information according to the underwriting information includes:

[0022] Extracting features from the underwriting information to obtain an underwriting feature set;

[0023] Analyzing the underwriting feature set using a risk assessment model to obtain a feature analysis result, wherein the risk assessment model is obtained by training and learning an initial risk assessment model based on data in the underwriting knowledge base;

[0024] Generate underwriting key point prompt information based on the feature analysis results.

[0025] Optionally, after generating the underwriting recommendation, the method further includes:

[0026] Obtaining underwriting decision results and feedback information for the underwriting recommendations;

[0027] The risk assessment model is optimized according to the underwriting decision result and the feedback information.

[0028] Optionally, after generating the underwriting recommendation, the method further includes:

[0029] The underwriting recommendation is presented in a visual form.

[0030] Optionally, the acquiring of underwriting inquiry information, parsing the underwriting inquiry information to obtain demand information corresponding to the underwriting inquiry information; searching for underwriting information corresponding to the demand information in an underwriting knowledge base, and generating underwriting key point prompt information according to the underwriting information; and generating underwriting suggestions according to the search enhancement-generated underwriting information and the underwriting key point prompt information, include:

[0031] Controlling the intelligent agent to obtain underwriting inquiry information, parsing the underwriting inquiry information, and obtaining demand information corresponding to the underwriting inquiry information;

[0032] Control the intelligent agent to search for underwriting information corresponding to the demand information in an underwriting knowledge base, and generate underwriting key point prompt information according to the underwriting information;

[0033] The intelligent agent is controlled to generate underwriting information and the underwriting key point prompt information according to the retrieval enhancement, and generate underwriting recommendations.

[0034] To achieve the above-mentioned purpose, a second embodiment of the present disclosure provides an underwriting recommendation generating device, including:

[0035] An information parsing unit, used to obtain underwriting inquiry information, parse the underwriting inquiry information, and obtain demand information corresponding to the underwriting inquiry information;

[0036] An information search unit, used to search for underwriting information corresponding to the demand information in an underwriting knowledge base, and generate underwriting key point prompt information according to the underwriting information, wherein the underwriting knowledge base includes retrieval enhancement to generate underwriting information;

[0037] A suggestion generating unit is used to generate underwriting information and the underwriting key point prompt information according to the search enhancement, and generate underwriting suggestions.

[0038] To achieve the above-mentioned purpose, a third aspect of the present disclosure provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0039] Memory stores computer-executable instructions;

[0040] The processor executes the computer-executable instructions stored in the memory to implement the method shown in any one of the first aspects above.

[0041] To achieve the above-mentioned objectives, an embodiment of the fourth aspect of the present disclosure proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the method shown in any one of the above-mentioned first aspects.

[0042] To achieve the above-mentioned purpose, the fifth aspect of the present disclosure provides a computer program product, including a computer program, which implements the method shown in any one of the above-mentioned first aspects when executed by a processor.

[0043] In summary, the method, apparatus, device, medium and program product provided by the present disclosure first obtains underwriting inquiry information, parses the underwriting inquiry information, obtains demand information corresponding to the underwriting inquiry information, searches the underwriting information corresponding to the demand information in the underwriting knowledge base, and generates underwriting key point prompt information based on the underwriting information; therefore, clear and specific underwriting key point prompts can be provided to underwriters to reduce the risk of human misjudgment. Next, by combining retrieval enhancement to generate underwriting information and underwriting key point prompt information to generate underwriting recommendations, personalized underwriting recommendations can be implemented to meet the specific needs and risk conditions of different policyholders. In addition, the degree of automation of the underwriting process can be improved, and human intervention can be reduced, thereby improving underwriting efficiency and accuracy.

[0044] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0046] Figure 1 A flowchart of a method for generating an underwriting recommendation provided by an embodiment of the present disclosure;

[0047] Figure 2 A schematic diagram of a model optimization process provided by an embodiment of the present disclosure;

[0048] Figure 3 A schematic diagram of the structure of an underwriting recommendation generating device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0049] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0050] The present disclosure is described in detail below with reference to specific embodiments.

[0051] In the first embodiment, if Figure 1 As shown, Figure 1This is a flow chart of a method for generating underwriting recommendations provided by an embodiment of the present disclosure. The method can be implemented by a computer program and can be run on a device for generating underwriting recommendations. The computer program can be integrated into an application or run as an independent tool application.

[0052] Among them, the underwriting suggestion generating device may be an electronic device with the function of generating underwriting suggestions.

[0053] Among them, the underwriting recommendation generating method can be executed by an electronic device.

[0054] For example, the underwriting recommendation generating method includes the following steps:

[0055] S101, obtaining underwriting inquiry information, parsing the underwriting inquiry information, and obtaining demand information corresponding to the underwriting inquiry information;

[0056] According to some embodiments, underwriting inquiry information refers to question information about underwriting input by a user, including but not limited to special contracts, notifications, underwriting records, non-standard conclusions, physical examination records, etc.

[0057] Among them, special agreements refer to certain specific insurance terms or agreements, which may affect the coverage, premiums or claims process of the policy; their specific meaning needs to be determined based on the specific insurance contract and underwriting situation.

[0058] Among them, disclosure is an important part of the life insurance underwriting process, which means that when purchasing insurance, the policyholder or the insured needs to truthfully disclose information related to insurance risks to the insurance company. This information usually includes personal health status, occupational risks, living habits, etc. The purpose of disclosure is to ensure that the insurance company can accurately assess the risk and thus formulate reasonable premiums and underwriting conditions. If the policyholder or the insured deliberately conceals or falsely states relevant information, it may cause the insurance contract to be invalid or disputes to arise during claims.

[0059] Among them, underwriting records are detailed records generated by insurance companies during the underwriting process, including personal information, health status, occupational risk assessment, family medical history, etc. of the policyholder or insured. These records are important basis for insurance companies to decide whether to provide insurance, how much the premium is, and the conditions for underwriting. The accuracy and completeness of underwriting records are crucial to both insurance companies and policyholders, as they directly affect the effectiveness of insurance contracts and claims processing.

[0060] Among them, non-standard conclusions refer to non-standard underwriting conditions or decisions given by insurance companies based on the specific circumstances of the policyholder or the insured in life insurance underwriting. This usually means that the risk status of the policyholder or the insured does not meet the standard underwriting conditions and requires additional consideration or restrictions. Non-standard conclusions may include surcharge underwriting (i.e., adding a certain fee on top of the standard premium), exclusions (i.e., excluding coverage for certain specific risks), deferred underwriting (i.e., temporarily not underwriting and making a decision after a period of observation) or rejection of insurance (i.e., not accepting insurance applications), etc. These conclusions are based on the insurance company's detailed assessment and analysis of the risk status of the policyholder or the insured.

[0061] In some embodiments, the underwriting inquiry information is parsed to extract key information from the underwriting inquiry information as demand information corresponding to the underwriting inquiry information, thereby clarifying the specific needs of underwriting assistance.

[0062] For example, when the underwriting inquiry information obtained is "A 45-year-old male with mild hypertension, can he be insured?", by parsing it, key information such as age (45 years old), gender (male), and health status (mild hypertension) can be extracted.

[0063] It is easy to understand that when underwriters raise questions, by intelligently analyzing the questions and accurately extracting key information, precise guidance can be provided for the subsequent generation of underwriting recommendations.

[0064] S102, searching for underwriting information corresponding to the demand information in an underwriting knowledge base, and generating underwriting key point prompt information according to the underwriting information;

[0065] According to some embodiments, the underwriting knowledge base refers to a collection of relevant information of the underwriting industry, including Retrieval Augmented Generation (RAG) underwriting information.

[0066] In some embodiments, RAG technology combines two functional components, "retrieval" and "generator", to handle complex information query and generation tasks. In the field of insurance underwriting, RAG technology can be used to generate RAG underwriting information, which includes multiple underwriting rules and can provide accurate and efficient decision support for underwriters.

[0067] According to some embodiments, underwriting information refers to knowledge related to demand information in an underwriting knowledge base.

[0068] In some embodiments, the underwriting key point prompt information refers to prompt information generated for underwriting information. The underwriting key point prompt information may include, for example, underwriting key points in key aspects such as special agreements, notifications, underwriting records, non-standard conclusions, and physical examination records.

[0069] For example, you can search for underwriting information related to "45-year-old male, mild hypertension" from the underwriting knowledge base, and generate underwriting key point prompts based on the underwriting information, such as "pay attention to cardiovascular risks" and "additional medical examinations may be required".

[0070] S103, generating underwriting information and key underwriting prompt information based on the search enhancement, and generating underwriting recommendations.

[0071] According to some embodiments, by combining RAG underwriting information and underwriting key point prompt information, specific and personalized underwriting recommendations can be generated, including but not limited to risk rating, recommended premium, insurance amount adjustment, premium adjustment recommendation, underwriting condition recommendation, additional physical examination items, etc., which can reduce the risk of human misjudgment.

[0072] In summary, the method provided in this embodiment first obtains underwriting inquiry information, parses the underwriting inquiry information, obtains demand information corresponding to the underwriting inquiry information, searches the underwriting information corresponding to the demand information in the underwriting knowledge base, and generates underwriting key point prompt information based on the underwriting information; therefore, clear and specific underwriting key point prompts can be provided to underwriters to reduce the risk of human misjudgment. Next, by combining retrieval enhancement to generate underwriting information and underwriting key point prompt information to generate underwriting recommendations, personalized underwriting recommendations can be implemented to meet the specific needs and risk conditions of different policyholders. In addition, the degree of automation of the underwriting process can be improved, and human intervention can be reduced, thereby improving underwriting efficiency and accuracy.

[0073] This embodiment also provides another method for generating underwriting recommendations, which can be executed by an electronic device.

[0074] For example, the underwriting recommendation generating method may include the following steps:

[0075] S201, obtaining search enhancement to generate underwriting information, and collecting policyholder related information from at least one data source;

[0076] According to some embodiments, the retrieved enhanced generated underwriting information may be, for example, information in an intelligent RAG underwriting manual created through RAG technology.

[0077] In some embodiments, the RAG underwriting manual can provide accurate and efficient decision support for underwriters by retrieving relevant underwriting regulations, cases and historical data, combined with the generation capabilities of large models.

[0078] In some embodiments, a large language model (LLM), referred to as a large model, refers to a machine learning model with large-scale parameters and complex computational structures. These models are usually built from deep neural networks and have billions or even hundreds of billions of parameters. The design purpose of large models is to improve the expressiveness and predictive performance of the model to handle more complex tasks and data. They learn complex patterns and features by training massive data, so they have more powerful generalization capabilities and can make accurate predictions for unseen data. In short, a large model is a machine learning model with large-scale parameters and complex structures.

[0079] According to some embodiments, the policyholder-related information includes, but is not limited to, special information, health notifications, historical underwriting records, non-standard conclusions, and physical examination records.

[0080] In some embodiments, data sources include but are not limited to medical institutions, public databases, third-party data service providers, policyholders, etc.

[0081] In some embodiments, methods for collecting information related to the insured include but are not limited to application programming interface (API) interface, data crawler, user input, etc.

[0082] For example, you can obtain the insured's physical examination report from a medical institution, the insured's credit record from a public database, and the insured's consumption habits and other data from a third-party service provider.

[0083] According to some embodiments, an AI agent may be used to obtain retrieval enhancement to generate underwriting information and collect information related to the policyholder from at least one data source.

[0084] In some embodiments, AI Agent is an intelligent entity that can perceive the environment, make decisions, and perform actions. AI Agent is different from traditional artificial intelligence. It has the ability to gradually complete a given goal by thinking independently and calling various tools. In simple terms, AI Agent can be regarded as an intelligent agent that can interact with the big model on our behalf, reducing the complexity of the user's interaction with the big model. In short, AI Agent is an intelligent entity that can interact with the big model on behalf of humans.

[0085] It is easy to understand that AI Agent can automatically capture and integrate policyholder-related information from multiple data sources to ensure the comprehensiveness and real-time nature of the data.

[0086] S202, integrating the underwriting information generated by the search enhancement and the information related to the policyholder to obtain integrated information;

[0087] According to some embodiments, a rich underwriting knowledge base may be constructed by integrating data such as the rules in the RAG underwriting manual, decision data of historical underwriting cases, and underwriting specifications commonly used in the industry.

[0088] For example, the RAG underwriting manual could include underwriting criteria for different age groups, genders and health conditions, which would be incorporated into the underwriting knowledge base.

[0089] S203, performing data preprocessing on the integrated information to obtain preprocessed integrated information, and constructing an underwriting knowledge base based on the preprocessed integrated information;

[0090] According to some embodiments, data preprocessing includes but is not limited to data cleaning, data labeling, data normalization, format conversion, standardization, etc., so as to improve data quality and ensure the comprehensiveness, real-time, accuracy and consistency of the data.

[0091] In some embodiments, data cleansing can remove duplicate, invalid or erroneous data, such as the insured's contact information, home address, etc.

[0092] In some embodiments, the data is standardized, for example, health status descriptions in different formats may be unified into standardized codes.

[0093] In some embodiments, various types of collected data are converted into a unified storage format to facilitate subsequent data analysis.

[0094] According to some embodiments, natural language processing technology may be used to perform data preprocessing on the integrated information, thereby improving the efficiency and accuracy of the data preprocessing.

[0095] In some embodiments, AI Agent may also use natural language processing technology to automatically pre-process and store the integrated information for subsequent analysis.

[0096] S204, obtaining underwriting inquiry information, parsing the underwriting inquiry information, obtaining demand information corresponding to the underwriting inquiry information, and recalling search information corresponding to the demand information from the underwriting knowledge base using search enhancement generation technology;

[0097] According to some embodiments, natural language processing technology can be used to parse underwriting inquiry information, thereby reducing human intervention and improving the speed and accuracy of underwriting decisions. At the same time, natural language processing technology can also more flexibly respond to various complex underwriting scenarios and improve the intelligence level of underwriting decisions.

[0098] In some embodiments, the intelligent agent can be controlled to obtain underwriting inquiry information, parse the underwriting inquiry information, and obtain demand information corresponding to the underwriting inquiry information; the intelligent agent can also be controlled to search for underwriting information corresponding to the demand information in the underwriting knowledge base, and generate underwriting key points prompt information based on the underwriting information.

[0099] For example, when an underwriter raises a question, the AI ​​agent can intelligently parse the question, accurately extract key information, initiate a search request based on the analysis requirements, and intelligently retrieve relevant underwriting information from the underwriting knowledge base to provide precise guidance for subsequent business enhancements.

[0100] According to some embodiments, the retrieval information corresponding to the demand information may be recalled from the underwriting knowledge base using the RAG technology through the RAG model.

[0101] In some embodiments, the retrieval information may include, for example, underwriting rules and cases, and the information may be sorted by relevance to ensure that the returned information is highly relevant to the user's question.

[0102] S205, performing intent recognition on the search information, and performing service enhancement on the search information according to the intent recognition result to obtain service enhancement information;

[0103] According to some embodiments, when performing intent recognition on retrieved information, the retrieved information can be interpreted in combination with the experience of underwriting experts and the contents of the RAG underwriting manual. For example, the usual underwriting strategy for the situation can be identified. The business enhancement information obtained by performing business enhancement on the retrieved information based on the intent recognition results can provide more comprehensive support for underwriting decisions.

[0104] S206, generating underwriting information corresponding to the demand information according to the search information and the business enhancement information;

[0105] According to some embodiments, the RAG model can be used to identify the intent of the retrieved information, and the service of the retrieved information can be enhanced based on the intent identification result to obtain the service enhancement information, and the retrieved information and the service enhancement information can be integrated to obtain the underwriting information corresponding to the demand information.

[0106] S207, extracting features from the underwriting information to obtain an underwriting feature set;

[0107] According to some embodiments, when extracting features from underwriting information, the extracted features include but are not limited to age, gender, health status, occupational risk, BMI index, etc.

[0108] In some embodiments, feature engineering can enhance the expressiveness of data and improve the predictive performance of the model.

[0109] S208, analyzing the underwriting feature set using a risk assessment model to obtain feature analysis results;

[0110] According to some embodiments, the risk assessment model may be obtained by training and learning the initial risk assessment model according to the data in the underwriting knowledge base. The risk assessment model may be, for example, an artificial intelligence (AI) model.

[0111] In some embodiments, the risk assessment model can identify risk patterns and trends in the data, assess the risk of the insured, and predict their future health status and possible compensation risks. The risk assessment model can also use big data analysis techniques, such as data mining, pattern recognition, association rule learning, etc., to conduct in-depth mining and analysis on pre-processed data, and use data mining techniques to discover potential associations between data, such as the risk relationship between certain health conditions and specific diseases. The risk assessment model can also use ensemble learning, deep learning and other technologies to improve the accuracy of the analysis.

[0112] In some embodiments, at least one machine learning algorithm may be used to train and learn the data. The machine learning algorithm includes but is not limited to logistic regression, support vector machine, random forest, neural network, etc.

[0113] It should be noted that through comprehensive analysis and mining of multi-source data in the underwriting knowledge base, the risk status of the insured can be more comprehensively assessed, providing a more adequate basis for underwriting decisions. This ability to integrate and comprehensively analyze multi-source data can enhance the comprehensiveness and accuracy of underwriting recommendations.

[0114] S209, generating underwriting key point prompt information according to the feature analysis result;

[0115] According to some embodiments, the feature analysis results obtained by using a trained risk assessment model to conduct a comprehensive analysis of the extracted underwriting features can be used to assess the insured's risk level and possible claims risk, automatically identify potential risk points, and provide a risk assessment report for underwriters. The information in the risk assessment report includes underwriting key points prompt information.

[0116] In some embodiments, the AI ​​Agent can comprehensively consider the retrieval information and business enhancement information generated by the RAG model to generate more accurate and personalized underwriting key points.

[0117] According to some embodiments, during the knowledge retrieval and business enhancement stages, full use can be made of the rules in the RAG underwriting manual and the experience of underwriting experts. This combination can make the generated underwriting key point prompt information more accurate and reliable, and can reduce the risk of human misjudgment.

[0118] It should be noted that by introducing big data analysis and AI technology, various data of policyholders can be automatically collected, integrated and pre-processed, including special information, health notifications, historical underwriting records, etc. After deep mining and processing by machine learning algorithms, these data can provide underwriters with comprehensive and accurate risk assessment reports. This intelligent decision support can significantly improve the efficiency and accuracy of underwriting decisions.

[0119] In addition, feature engineering can be used to conduct personalized risk assessment based on the specific circumstances of each policyholder. By comprehensively considering factors such as age, gender, health status, occupational risk, etc., it is possible to generate risk ratings and premium adjustment recommendations for individuals. This personalized risk assessment not only meets the specific needs of different policyholders, but also enhances the scientificity and rationality of underwriting decisions.

[0120] S210, generating underwriting information and underwriting key point prompt information according to the search enhancement, generating underwriting suggestions, and displaying the underwriting suggestions in a visual form;

[0121] According to some embodiments, the intelligent agent can be controlled to generate underwriting information and underwriting key point prompt information based on the retrieval enhancement, and generate underwriting recommendations.

[0122] For example, AI Agent can generate more accurate and personalized underwriting recommendations based on underwriting key points prompts combined with the RAG underwriting manual and business rules.

[0123] In some implementations, the underwriting recommendation is clear and specific, making it easy for users to understand and implement. For example, the underwriting recommendation may be "after adding specific physical examination items, coverage can be considered, but the premium will need to increase by 10%."

[0124] It should be noted that, based on risk assessment, specific underwriting recommendations are automatically generated, including underwriting conditions, premium adjustments, and recommended physical examination items. These recommendations are generated through intelligent analysis and are based on the rules of the RAG underwriting manual and the experience of underwriting experts. This can ensure the accuracy and reliability of the recommendations and reduce the risk of human misjudgment. Underwriters can make decisions quickly based on these recommendations, improving the efficiency and effectiveness of the underwriting process.

[0125] According to some embodiments, when the underwriting recommendations are presented in a visual form, the visual form includes but is not limited to charts, texts, reports, etc., which can be presented to the underwriters clearly and intuitively, facilitating the underwriters to quickly understand and make decisions.

[0126] In some embodiments, when displaying underwriting recommendations, advanced analytical information such as historical data comparisons and industry benchmark comparisons may also be displayed.

[0127] S211, obtaining the underwriting decision results and feedback information input for the underwriting suggestion, and optimizing the risk assessment model based on the underwriting decision results and feedback information.

[0128] According to some embodiments, the feedback information is used to indicate the accuracy of the underwriting suggestion. For example, the feedback information may be "the suggestion is too conservative", "the suggestion is very accurate", etc.

[0129] In some embodiments, after the underwriter makes a final decision based on the underwriting recommendation, the AI ​​agent can automatically collect the underwriting decision results and feedback information of the underwriter, and automatically analyze the underwriting decision results and feedback information, so as to continuously optimize the risk assessment model and RAG model to make them closer to actual underwriting needs and standards, gradually improve the accuracy of risk assessment, improve the adaptability, accuracy and effectiveness of underwriting recommendation generation, and improve the intelligence level of underwriting recommendation generation.

[0130] In some embodiments, the optimization of the risk assessment model and the RAG model includes but is not limited to knowledge optimization, Prompt optimization, model parameter adjustment, etc.

[0131] According to some embodiments, the performance of the AI ​​Agent and the RAG model may also be monitored and evaluated in real time to ensure the stability and reliability of the generation of underwriting recommendations.

[0132] It should be noted that by introducing the intelligent application of AI agents in multiple links such as data collection, preprocessing, demand analysis, knowledge retrieval, underwriting recommendation generation and feedback optimization, the automation level and accuracy of the underwriting process can be further improved, and various complex situations can be responded to more intelligently, achieving more personalized and accurate underwriting services.

[0133] According to some embodiments, Figure 2 A schematic diagram of a model optimization process provided by an embodiment of the present disclosure. Figure 2As shown in the figure, first, the scene data is collected, that is, knowledge reserves and sorting, including FAQs, special contracts, notifications, underwriting records, non-standard conclusions, physical examination records, document manuals, dialogue knowledge, etc. These data are the basis for model training and application. Then, the collected data is integrated with underwriting dictionaries, underwriting key points, underwriting summaries, underwriting conclusions, RAG underwriting manuals and other data. After cleaning and annotating the integrated data, it is input into the supervised learning model BERT for preliminary processing. Subsequently, the process turns to model training and application; among them, the RAG model is first pre-trained, and then fine-tuned according to specific tasks to optimize its performance; in the inference stage, the RAG model can generate accurate answers based on the input data; among them, the semi-supervised learning model is used for data enhancement, risk assessment and consistency prediction. Then, in the model iteration stage, the performance and accuracy of the model are continuously improved through knowledge optimization, prompt optimization, RAG model optimization, large model optimization and other means; at the same time, the model is analyzed for error data in order to make targeted improvements. Finally, the trained and optimized model can be applied to the underwriting field to realize functions such as knowledge storage, reasoning, and generation of troubleshooting conclusions. The entire process can reflect the powerful application potential of the model in underwriting scenarios.

[0134] It should be noted that the underwriting dictionary and underwriting key points play a key role in the model training and optimization process. The underwriting dictionary provides the basis for knowledge representation and matching in the underwriting process, while the underwriting key points point out the key points that need special attention in the underwriting process. In addition, professional data annotation is also an indispensable part of model training, which ensures that the model can accurately understand and process data in professional fields.

[0135] For example, taking the health insurance underwriting scenario, a 45-year-old male policyholder applies to purchase a health insurance policy. He fills in the application form that he has a history of mild hypertension and provides a recent physical examination report. When the insurance company uses the method provided in this embodiment to conduct underwriting assessment, first, the AI ​​Agent automatically obtains the insured's physical examination report from the medical institution, cleans and formats the data, and extracts key information such as age, gender, blood pressure value, etc.; then, the underwriter asks a question: "A 45-year-old male with mild hypertension, can it be covered?" The AI ​​Agent parses the question and clarifies the underwriting needs; then, the RAG model is used to retrieve relevant rules and cases from the underwriting knowledge base, and business enhancement is performed in combination with the experience of underwriting experts to identify the underwriting strategy in this case; then, the AI ​​Agent generates underwriting key points, such as "Attention should be paid to cardiovascular risks, and further evaluation of hypertension control is recommended"; then, combined with the RAG underwriting manual, the underwriting conclusion and suggestion are output: "After providing proof of stable blood pressure control in the past six months, underwriting can be considered, but the premium needs to be increased by 5%"; finally, the underwriter makes a decision based on the generated underwriting suggestion, and the AI ​​Agent collects feedback data for model optimization to improve the accuracy of subsequent underwriting.

[0136] For example, taking the occupational risk insurance underwriting scenario as an example, a worker engaged in high-altitude operations applies to purchase occupational risk insurance. This occupation has a high risk. When the insurance company uses the method provided in this embodiment to conduct a detailed underwriting assessment, first, the AI ​​Agent obtains the insured's occupational information, historical claims records and other data from the third-party data service provider, and performs cleaning and formatting processing; then, the underwriter asks a question: "How to assess the occupational risks and determine the premiums of workers working at heights?" The AI ​​Agent parses the question and clarifies the underwriting needs; then, the RAG model is used to retrieve risk assessment standards and historical claims cases related to high-altitude operations from the underwriting knowledge base, and business enhancement is performed in combination with the experience of underwriting experts; then, the AI ​​Agent generates underwriting key points, such as "the insured's health status, work experience, safety measures, etc. need to be assessed"; then, after comprehensively assessing the insured's various risk factors, the underwriting conclusion and suggestions are output: "In view of the high risk of high-altitude operations, it is recommended to add specific physical examination items (such as electrocardiograms and pulmonary function tests), and the premium will be increased by 20%"; finally, the underwriter makes a decision based on the generated underwriting suggestions, and the AI ​​Agent collects feedback data for model optimization to improve the accuracy of subsequent underwriting.

[0137] For example, taking the personalized premium pricing scenario as an example, a young female policyholder applies to purchase life insurance, and she hopes to obtain personalized premium pricing based on her health status and living habits. When the insurance company uses the method provided in this embodiment to conduct a detailed underwriting assessment, first, the AI ​​Agent collects the insured's age, gender, weight, exercise habits, family medical history and other data through user input and third-party data sources, and performs cleaning and formatting processing; then, the insured asks a question: "I hope to get personalized premium pricing based on my personal situation." The AI ​​Agent parses the question and clarifies the underwriting needs; then, the system uses big data analysis technology and machine learning algorithms to conduct in-depth analysis of the collected data to identify the insured's risk characteristics and premium sensitive factors; then, the AI ​​Agent generates personalized underwriting key points, such as "The insured is young, has no family medical history, but is slightly overweight, and it is recommended to increase exercise"; then, combined with the RAG underwriting manual and business rules, the insured is provided with personalized premium pricing suggestions: "Considering your age and health status, it is recommended that the premium float -5% to +0% based on the standard rate, depending on your subsequent health improvement"; finally, the insured makes a decision based on the generated underwriting suggestions and can choose whether to accept the premium pricing. The AI ​​Agent collects feedback data for model optimization to improve the accuracy and satisfaction of personalized premium pricing in the future.

[0138] In summary, the method provided in this embodiment is different from the traditional underwriting process that relies on manual labor. It realizes the intelligence of the underwriting process by utilizing big data and AI technology, especially AI Agent and RAG model. This deep integration not only improves the efficiency and accuracy of data processing, but also makes the underwriting process more automated and standardized, and can more accurately assess risks, reduce manual intervention, and reduce human misjudgment, thereby improving the efficiency and accuracy of underwriting; among them, through automated data collection, deep analysis and intelligent suggestion generation, it can greatly reduce manual processing time and improve the efficiency and accuracy of the underwriting process. Secondly, risk assessment and premium pricing can be carried out according to the specific circumstances of each policyholder, to achieve personalized underwriting suggestions, improve customer satisfaction, and enhance the scientificity and rationality of underwriting decisions. In addition, the model is continuously optimized through the feedback mechanism, so that the model can continue to learn and improve to adapt to the ever-changing market environment and policyholder needs, making the generation of underwriting suggestions more intelligent and accurate.

[0139] In order to implement the above embodiments, the present disclosure also proposes an underwriting recommendation generating device.

[0140] like Figure 3 As shown, the underwriting suggestion generating device 300 includes:

[0141] The information parsing unit 301 is used to obtain the underwriting inquiry information, parse the underwriting inquiry information, and obtain the demand information corresponding to the underwriting inquiry information;

[0142] The information search unit 302 is used to search the underwriting information corresponding to the demand information in the underwriting knowledge base, and generate the underwriting key point prompt information according to the underwriting information, wherein the underwriting knowledge base includes the search enhancement to generate the underwriting information;

[0143] The suggestion generating unit 303 is used to generate underwriting information and underwriting key point prompt information according to the search enhancement, and generate underwriting suggestions.

[0144] Optionally, before searching the underwriting information corresponding to the demand information in the underwriting knowledge base, the information searching unit 302 is further configured to:

[0145] Acquire search enhancement to generate underwriting information, and collect policyholder related information from at least one data source;

[0146] Performing data integration on the underwriting information generated by the search enhancement and the relevant information of the policyholder to obtain integrated information;

[0147] The integrated information is preprocessed to obtain preprocessed integrated information, and an underwriting knowledge base is constructed based on the preprocessed integrated information.

[0148] Optionally, when the information searching unit 302 is used to search for the underwriting information corresponding to the demand information in the underwriting knowledge base, it is specifically used to:

[0149] Recall the search information corresponding to the demand information from the underwriting knowledge base using the search enhancement generation technology;

[0150] Performing intent recognition on the searched information, and performing business enhancement on the searched information according to the intent recognition result to obtain business enhancement information;

[0151] Generate underwriting information corresponding to the demand information based on the search information and business enhancement information.

[0152] Optionally, when the information search unit 302 is used to generate the underwriting key point prompt information according to the underwriting information, it is specifically used to:

[0153] Extract features from the underwriting information to obtain an underwriting feature set;

[0154] The risk assessment model is used to analyze the underwriting feature set to obtain feature analysis results, wherein the risk assessment model is obtained by training and learning the initial risk assessment model based on the data in the underwriting knowledge base;

[0155] Generate underwriting key point prompt information based on feature analysis results.

[0156] Optionally, after generating the underwriting suggestion, the suggestion generating unit 303 is further configured to:

[0157] Obtain underwriting decision results and feedback information for underwriting recommendation input;

[0158] Optimize the risk assessment model based on underwriting decision results and feedback information.

[0159] Optionally, after generating the underwriting suggestion, the suggestion generating unit 303 is further configured to:

[0160] Present underwriting recommendations in a visual format.

[0161] Optionally, the information parsing unit 301 is specifically used to control the intelligent agent to obtain the underwriting inquiry information, parse the underwriting inquiry information, and obtain the demand information corresponding to the underwriting inquiry information;

[0162] The information search unit 302 is specifically used to control the intelligent agent to search for the underwriting information corresponding to the demand information in the underwriting knowledge base, and generate the underwriting key point prompt information according to the underwriting information;

[0163] The suggestion generation unit 303 is specifically used to control the intelligent agent to generate underwriting information and underwriting key point prompt information according to the retrieval enhancement, and generate underwriting suggestions.

[0164] It should be noted that the aforementioned explanation of the embodiment of the underwriting recommendation generating method is also applicable to the underwriting recommendation generating device of this embodiment, and will not be repeated here.

[0165] In summary, the device provided by the embodiment of the present disclosure, first, obtains underwriting inquiry information, parses the underwriting inquiry information, obtains demand information corresponding to the underwriting inquiry information, searches the underwriting information corresponding to the demand information in the underwriting knowledge base, and generates underwriting key point prompt information based on the underwriting information; therefore, it can provide clear and specific underwriting key point prompts for underwriting personnel to reduce the risk of human misjudgment. Next, by combining retrieval enhancement to generate underwriting information and underwriting key point prompt information to generate underwriting suggestions, personalized underwriting suggestions can be achieved to meet the specific needs and risk conditions of different policyholders. In addition, the degree of automation of the underwriting process can be improved, and manual intervention can be reduced, thereby improving underwriting efficiency and accuracy. Secondly, the units work together to complete the task of intelligent underwriting assistance. This modular design makes the device more flexible and scalable, and facilitates subsequent functional upgrades and optimizations.

[0166] In order to implement the above embodiments, the present disclosure also proposes an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0167] In order to implement the above embodiments, the present disclosure further proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0168] In order to implement the above embodiments, the present disclosure also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0169] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this disclosure shall comply with the relevant laws and regulations and shall not violate public order and good morals.

[0170] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0171] The present disclosure anticipates providing implementation schemes for users to selectively block the use or access of personal information data. That is, the present disclosure anticipates providing hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, risks can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0172] The acquisition, transmission, storage, use, and processing of data in the technical solution disclosed in this disclosure are in compliance with the relevant provisions of national laws and regulations.

[0173] It should be noted that in the embodiments of the present disclosure, certain software, components, models and other existing solutions in the industry may be mentioned, which should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0174] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0175] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0176] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present disclosure belong.

[0177] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or otherwise processing in a suitable manner if necessary, and then stored in a computer memory.

[0178] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0179] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0180] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0181] The storage medium mentioned above may be a read-only memory, a disk or an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present disclosure. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present disclosure.

Claims

1. A method for generating an underwriting recommendation, characterized in that: include: Acquire underwriting inquiry information, parse the underwriting inquiry information, and obtain demand information corresponding to the underwriting inquiry information; Searching for underwriting information corresponding to the demand information in an underwriting knowledge base, and generating underwriting key point prompt information according to the underwriting information, wherein the underwriting knowledge base includes retrieval enhancement to generate underwriting information; The underwriting information and the underwriting key point prompt information are generated based on the search enhancement, and an underwriting recommendation is generated.

2. The method according to claim 1, characterized in that: Before searching the underwriting information corresponding to the demand information in the underwriting knowledge base, the method further includes: Acquire search enhancement to generate underwriting information, and collect policyholder related information from at least one data source; Performing data integration on the underwriting information generated by the search enhancement and the policyholder related information to obtain integrated information; The integrated information is subjected to data preprocessing to obtain preprocessed integrated information, and an underwriting knowledge base is constructed based on the preprocessed integrated information.

3. The method according to claim 1, characterized in that The searching the underwriting information corresponding to the demand information in the underwriting knowledge base includes: Recalling the search information corresponding to the demand information from the underwriting knowledge base using the search enhancement generation technology; Performing intent recognition on the search information, and performing service enhancement on the search information according to the intent recognition result to obtain service enhancement information; Generate underwriting information corresponding to the demand information based on the search information and the service enhancement information.

4. The method according to claim 1, characterized in that: The generating of the underwriting key point prompt information according to the underwriting information includes: Extracting features from the underwriting information to obtain an underwriting feature set; Analyzing the underwriting feature set using a risk assessment model to obtain a feature analysis result, wherein the risk assessment model is obtained by training and learning an initial risk assessment model based on data in the underwriting knowledge base; Generate underwriting key point prompt information based on the feature analysis results.

5. The method according to claim 4, characterized in that After generating the underwriting recommendation, the method further includes: Obtaining underwriting decision results and feedback information for the underwriting recommendations; The risk assessment model is optimized according to the underwriting decision result and the feedback information.

6. The method according to claim 1, characterized in that After generating the underwriting recommendation, the method further includes: The underwriting recommendation is presented in a visual form.

7. The method according to claim 1, characterized in that The underwriting inquiry information is obtained, the underwriting inquiry information is parsed to obtain the demand information corresponding to the underwriting inquiry information; the underwriting information corresponding to the demand information is searched in the underwriting knowledge base, and the underwriting key point prompt information is generated according to the underwriting information; Generating underwriting information and underwriting key point prompt information according to the search enhancement, and generating underwriting suggestions, including: Controlling the intelligent agent to obtain underwriting inquiry information, parsing the underwriting inquiry information, and obtaining demand information corresponding to the underwriting inquiry information; Control the intelligent agent to search for underwriting information corresponding to the demand information in an underwriting knowledge base, and generate underwriting key point prompt information according to the underwriting information; The intelligent agent is controlled to generate underwriting information and the underwriting key point prompt information according to the retrieval enhancement, and generate underwriting recommendations.

8. A device for generating underwriting recommendations, characterized in that: include: An information parsing unit, used to obtain underwriting inquiry information, parse the underwriting inquiry information, and obtain demand information corresponding to the underwriting inquiry information; An information search unit, used to search for underwriting information corresponding to the demand information in an underwriting knowledge base, and generate underwriting key point prompt information according to the underwriting information, wherein the underwriting knowledge base includes retrieval enhancement to generate underwriting information; A suggestion generating unit is used to generate underwriting information and the underwriting key point prompt information according to the search enhancement, and generate underwriting suggestions.

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

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

11. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 7 when being executed by a processor.

Images