Policy complaint early warning methods, devices, equipment, and media

Abstract

This application relates to the field of financial technology and provides a method, device, equipment, and medium for early warning of policy complaints. The method uses a complaint warning model to predict complaints for new insurance contracts before and after policy underwriting, thereby identifying high-risk policies and achieving early warning of policy complaints. On the one hand, by providing early warning of complaints for new insurance contracts, and since new policy underwriting is the source of insurance policies, improving the sales quality of new policies can fundamentally resolve and reduce policy complaints, further protecting customer rights and maintaining the reputation of the company and the industry. On the other hand, by establishing a model identification mechanism before and after underwriting, the policy complaint warning mechanism can be improved, forming a normalized complaint reduction mechanism.

Description

Technical Field

[0001] This application relates to the field of financial technology, and in particular to a method, device, equipment, and medium for issuing early warnings of policy complaints. Background Technology

[0002] In response to government regulatory policies on insurance company complaints and to reduce the negative impact of policy complaints on the company's reputation, insurance companies urgently need to implement policy complaint early warning systems to proactively resolve and reduce policy complaints.

[0003] The current policy complaint early warning system lacks a standardized early warning mechanism and relies mainly on subjective human judgment, lacking objective judgment standards. As a result, the efficiency and accuracy of predictions are difficult to guarantee, and the overall early warning effect is poor. Summary of the Invention

[0004] The main objective of this application is to propose a method, device, equipment, and medium for early warning of policy complaints, which can effectively realize early warning of policy complaints and improve the early warning mechanism for policy complaints.

[0005] To achieve the above objectives, a first aspect of this application proposes a policy complaint early warning method, the method comprising:

[0006] Before the new insurance contract is underwritten, obtain the target policy information, first agent information and first policyholder information of the new insurance contract from the policy underwriting platform;

[0007] The first complaint risk value is obtained by using the trained first complaint warning model based on the target policy information, the first agent information and the first policyholder information. The first complaint risk value represents the probability that a complaint will occur within a first preset time period after the new contract policy is underwritten.

[0008] The new insurance policy with the first complaint risk value being greater than or equal to the first preset threshold is sent as a high complaint risk policy to the policy underwriting platform.

[0009] At the first point after the new insurance contract is underwritten, the information of the second agent and the second policyholder of the new insurance contract is obtained from the policy underwriting platform.

[0010] The trained second complaint warning model is used to obtain a second complaint risk value based on the target policy information, the second agent information, and the second policyholder information. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point.

[0011] The new insurance policy with a second complaint risk value greater than or equal to the second preset threshold is sent to the policy underwriting platform as a high complaint risk policy.

[0012] According to the policy complaint early warning method provided in some embodiments of this application, the first complaint early warning model is trained through the following steps:

[0013] Obtain a first historical policy sample set, which includes a first positive sample and a first negative sample. The first positive sample is a policy that received a complaint within the first preset time period after the underwriting was completed, and the first negative sample is a policy that did not receive a complaint within the first preset time period after the underwriting was completed.

[0014] Feature extraction is performed on the policy information, agent information, and policyholder information in the first positive sample and the first negative sample to obtain the first training feature information;

[0015] The training objective is to predict whether a complaint will arise within a first preset time period after the policy is underwritten. Based on the first training feature information, a preset first complaint warning model is trained to update the model parameters of the first complaint warning model until the model performance of the first complaint warning model reaches the preset first training termination condition.

[0016] According to the policy complaint early warning method provided in some embodiments of this application, the second complaint early warning model is trained through the following steps:

[0017] Obtain second historical policy sample data, which includes second positive samples and second negative samples. The second positive samples are policies for which complaints occurred within the second preset time period after the first time point, and the second negative samples are policies for which no complaints occurred within the second preset time period after the first time point.

[0018] Feature extraction is performed on the policy information, agent information, and policyholder information in the second positive sample and the second negative sample to obtain the second training feature information;

[0019] The training objective is to predict whether a complaint will arise in the second preset time period after the first time point. Based on the second training feature information, a preset second complaint warning model is trained to update the model parameters of the second complaint warning model until the model performance of the second complaint warning model reaches the preset second training termination condition.

[0020] According to some embodiments of the present application, the policy complaint early warning method includes agent behavior information in both the first agent information and the second agent information. The policy complaint early warning method obtains the first agent information of the new insurance policy through the following steps:

[0021] The first agent information for the new insurance contract is obtained from the agent information database preset in the policy underwriting platform.

[0022] Before obtaining the second agent information and the second policyholder information of the new insurance contract from the policy underwriting platform, the method further includes:

[0023] Collect information on changes in the agent's first behavior during the period from the completion of underwriting to the first point in time for the new insurance policy;

[0024] Update the agent behavior information in the agent information database according to the first behavior change information;

[0025] The policy complaint early warning method obtains the second agent information for the new insurance policy through the following steps:

[0026] Retrieve the second agent information for the new insurance policy from the updated agent information database.

[0027] According to some embodiments of the policy complaint early warning method provided in this application, both the first policyholder information and the second policyholder information include policyholder behavior information. The policy complaint early warning method obtains the first policyholder information of the new contract policy through the following steps:

[0028] The first policyholder information of the new contract policy is obtained from the policyholder information database preset in the policy underwriting platform.

[0029] Before obtaining the second policyholder information and the second policyholder data of the new insurance contract from the policy underwriting platform, the method further includes:

[0030] Collect information on changes in the first behavior of the policyholder of the new insurance contract from the completion of underwriting to the first point in time;

[0031] Update the policyholder behavior information in the policyholder information database according to the first behavior change information;

[0032] The policy complaint early warning method obtains the second policyholder information of the new insurance contract through the following steps:

[0033] The second policyholder information for the new insurance policy is obtained from the updated policyholder information database.

[0034] According to some embodiments of this application, the policy complaint early warning method further includes:

[0035] At the second point in time after the underwriting is completed, the information of the third agent and the third policyholder of the new contract policy is obtained from the policy underwriting platform. The second point in time is the time point after the first point in time.

[0036] The trained third complaint warning model is used to obtain a third complaint risk value based on the target policy information, the third agent information, and the third policyholder information. The third complaint risk value represents the probability that a complaint will occur in the new contract policy within a third preset time period after the second time point.

[0037] New insurance policies with a third complaint risk value greater than or equal to a third preset threshold are sent to the policy underwriting platform as high complaint risk policies.

[0038] According to some embodiments of this application, the method for early warning of policy complaints includes obtaining a first complaint risk value using a trained first complaint warning model based on the target policy information, the first agent information, and the first policyholder information, including:

[0039] Feature extraction is performed on the target policy information, the first agent information, and the first policyholder information to obtain the first target feature information;

[0040] The first target feature information is input into the trained first complaint warning model so that the first complaint risk value is output through the first complaint warning model;

[0041] The process of using a trained second complaint warning model to obtain a second complaint risk value based on the target policy information, the second agent information, and the second policyholder information includes:

[0042] Feature extraction is performed on the target policy information, the second agent information, and the second policyholder information to obtain the second target feature information;

[0043] The second target feature information is input into the trained second complaint warning model, so that the second complaint risk value is output through the second complaint warning model.

[0044] To achieve the above objectives, a second aspect of this application provides a policy complaint early warning device, the device comprising:

[0045] The first acquisition module is used to acquire the target policy information, first agent information and first policyholder information of the new contract policy from the policy underwriting platform before the new contract policy is underwritten.

[0046] The first early warning module is used to obtain a first complaint risk value based on the target policy information, the first agent information, and the first policyholder information using a trained first complaint early warning model. The first complaint risk value represents the probability that a complaint will occur within a first preset time period after the new contract policy is underwritten.

[0047] The first sending module is used to send the new insurance contract with the first complaint risk value greater than or equal to the first preset threshold as a high complaint risk policy to the policy underwriting platform.

[0048] The second acquisition module is used to acquire the second agent information and the second policyholder information of the new contract policy from the policy underwriting platform at the first time point after the new contract policy is underwritten.

[0049] The second early warning module is used to obtain a second complaint risk value based on the target policy information, the second agent information, and the second policyholder information using a trained second complaint early warning model. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point.

[0050] The second sending module is used to send the new contract insurance policy with the second complaint risk value greater than or equal to the second preset threshold as a high complaint risk policy to the policy underwriting platform.

[0051] To achieve the above objectives, a third aspect of this application provides an electronic device, the electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the method described in the first aspect.

[0052] To achieve the above objectives, a fourth aspect of the present application provides a storage medium, which is a computer-readable storage medium for computer-readable storage, wherein the storage medium stores one or more computer programs that can be executed by one or more processors to implement the method described in the first aspect.

[0053] This application proposes a policy complaint early warning method, device, electronic device, and computer-readable storage medium. The method, before a new insurance contract is underwritten, obtains the target policy information, first agent information, and first policyholder information from the policy underwriting platform. It then uses a first complaint risk early warning model to obtain a first complaint risk value based on the target policy information, first agent information, and first policyholder information. This first complaint risk value represents the probability of a complaint occurring within a first preset time period after the new contract is underwritten. Furthermore, it identifies new contracts with a first complaint risk value greater than or equal to a first preset threshold. The policy is sent to the policy underwriting platform as a high-complaint-risk policy. At the first point after the new contract policy is underwritten, the second agent information and the second policyholder information of the new contract policy are obtained from the policy underwriting platform. Using a trained second complaint early warning model, the second complaint risk value is obtained based on the target policy, the second agent information, and the second policyholder information. The second complaint risk value represents the probability value of a complaint occurring in the second preset time period after the first time point. Then, new contract policies with a second complaint risk value greater than or equal to the second preset threshold are sent to the policy underwriting platform as high-complaint-risk policies. This application utilizes a complaint early warning model to predict complaints for new insurance contracts before and after policy underwriting, thereby identifying high-complaint-risk policies and achieving policy complaint early warning. On the one hand, by providing complaint early warnings for new insurance contracts, and since new policy underwriting is the source of insurance policies, improving the sales quality of new policies can fundamentally resolve and reduce policy complaints, further protecting customer rights and maintaining the company's and industry's reputation. On the other hand, by establishing a model identification mechanism before and after underwriting, the policy complaint early warning mechanism can be improved, forming a normalized complaint reduction mechanism. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating a policy complaint early warning method provided in an embodiment of this application;

[0055] Figure 2 This is a flowchart illustrating the training process of the first complaint warning model provided in this embodiment of the application;

[0056] Figure 3 This is a flowchart illustrating the training process of the second complaint warning model provided in this embodiment of the application;

[0057] Figure 4 This is a flowchart illustrating a policy complaint early warning method provided in another embodiment of this application;

[0058] Figure 5 yes Figure 1 A flowchart illustrating the sub-steps of step S120;

[0059] Figure 6 yes Figure 1A flowchart illustrating the sub-steps of step S150.

[0060] Figure 7 This is a flowchart illustrating a policy complaint early warning method provided in an embodiment of this application;

[0061] Figure 8 This is a schematic diagram of the structure of a policy complaint early warning device provided in an embodiment of this application;

[0062] Figure 9 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0065] First, let's analyze some of the terms used in this application:

[0066] Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0067] In response to government regulatory policies on insurance company complaints and to reduce the negative impact of policy complaints on the company's reputation, insurance companies urgently need to implement policy complaint early warning systems to proactively resolve and reduce policy complaints.

[0068] The current policy complaint early warning system lacks a standardized early warning mechanism and relies mainly on subjective human judgment, lacking objective judgment standards. As a result, the efficiency and accuracy of predictions are difficult to guarantee, and the overall early warning effect is poor.

[0069] Based on this, the embodiments of this application provide a policy complaint early warning method, device, electronic device and computer-readable storage medium, which can effectively realize policy complaint early warning and improve the policy complaint early warning mechanism.

[0070] First, the policy complaint early warning method in the embodiments of this application is described:

[0071] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0072] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0073] The policy complaint early warning method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the policy complaint early warning method, but is not limited to the above forms.

[0074] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0075] Please see Figure 1 , Figure 1 This document illustrates a flowchart of a policy complaint early warning method provided in an embodiment of this application. Figure 1 As shown, the policy complaint warning method includes, but is not limited to, steps S110 to S160.

[0076] Step S110: Before the new insurance contract is underwritten, obtain the target policy information, first agent information, and first policyholder information of the new insurance contract from the policy underwriting platform.

[0077] Step S120: Using the trained first complaint warning model, a first complaint risk value is obtained based on the target policy information, the first agent information, and the first policyholder information. The first complaint risk value represents the probability that a complaint will occur within a first preset time period after the new contract policy is underwritten.

[0078] Step S130: Send the new insurance policy with the first complaint risk value greater than or equal to the first preset threshold as a high complaint risk policy to the policy underwriting platform.

[0079] Step S140: At the first point after the new insurance contract is underwritten, obtain the information of the second agent and the second policyholder of the new insurance contract from the policy underwriting platform.

[0080] Step S150: Using the trained second complaint warning model, a second complaint risk value is obtained based on the target policy information, the second agent information, and the second policyholder information. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point.

[0081] Step S160: Send the new contract policy with the second complaint risk value greater than or equal to the second preset threshold as a high complaint risk policy to the policy underwriting platform.

[0082] Steps S110 to S160 in this embodiment of the application are as follows: Before the new contract policy is underwritten, the target policy information, first agent information, and first policyholder information of the new contract policy are obtained from the policy underwriting platform. A first complaint risk warning model is used to obtain a first complaint risk value based on the target policy information, first agent information, and first policyholder information. The first complaint risk value represents the probability value of a complaint occurring within a first preset time period after the new contract policy is underwritten. New contract policies with a first complaint risk value greater than or equal to a first preset threshold are sent to the policy underwriting platform as high complaint risk policies. At the first time point after the new contract policy is underwritten, the second agent information and second policyholder information of the new contract policy are obtained from the policy underwriting platform. A trained second complaint warning model is used to obtain a second complaint risk value based on the target policy, second agent information, and second policyholder information. The second complaint risk value represents the probability value of a complaint occurring within a second preset time period after the first time point. New contract policies with a second complaint risk value greater than or equal to a second preset threshold are then sent to the policy underwriting platform as high complaint risk policies. This application utilizes a complaint early warning model to predict complaints for new insurance contracts before and after policy underwriting, thereby identifying high-complaint-risk policies and achieving policy complaint early warning. On the one hand, by providing complaint early warnings for new insurance contracts, and since new policy underwriting is the source of insurance policies, improving the sales quality of new policies can fundamentally resolve and reduce policy complaints, further protecting customer rights and maintaining the company's and industry's reputation. On the other hand, by establishing a model identification mechanism before and after underwriting, the policy complaint early warning mechanism can be improved, forming a normalized complaint reduction mechanism.

[0083] The above is a general description of steps S110 to S160. Steps S110 to S160 will be described in detail below.

[0084] In step S110, before the new insurance contract is underwritten, the target policy information, first agent information, and first policyholder information of the new insurance contract are obtained from the policy underwriting platform.

[0085] Understandably, the target policy information, first agent information, and first policyholder information of a new contract policy can be obtained from the policy information database, agent information database, and policyholder information database pre-established in the policy underwriting platform.

[0086] It should be understood that policy information may include the type of insurance, the insurance period, and the insurance premium; agent information may include years of service, job level, and work quality information (such as complaints received and the renewal rate of clients under the agent's name); policyholder information may include basic information (such as gender, age, marital status, and income), coverage information (such as historical policy information), information on policies under their name (such as the time since the most recent policy and the maximum sum insured), information on complaints (such as whether they have received calls or expressed dissatisfaction on the insurance application in the past 7 / 30 / 90... days, or whether they have received calls or expressed their intention to cancel the policy in the past 7 / 30 / 90... days), and information on financial difficulties (such as whether they have experienced financial difficulties in the past 7 / 30 / 90... days).

[0087] It should be noted that agent information can be divided into basic agent information and agent behavioral information. Basic agent information may include the agent's years of service and job title, while agent behavioral information may include information about the agent's work quality, etc. Similarly, policyholder information can be divided into basic policyholder information and policyholder behavioral information. Basic policyholder information may include the policyholder's basic information, coverage information, and policy information under their name, while policyholder behavioral information may include information about complaints, financial difficulties, etc.

[0088] In steps S120 and S130, a first complaint risk value is obtained using the trained first complaint warning model based on the target policy information, the first agent information, and the first policyholder information. The first complaint risk value represents the probability that a complaint will occur within a first preset time period after the new contract policy is underwritten. New contract policies with a first complaint risk value greater than or equal to a first preset threshold are sent to the policy underwriting platform as high complaint risk policies.

[0089] It should be understood that a well-trained first complaint warning model uses information from three dimensions—target policy information, first agent information, and first policyholder information—as indicators to identify whether a new contract policy is a high-complaint-risk policy. This helps improve the accuracy of policy complaint warnings. A high-complaint-risk policy can be defined as a policy in which the probability of a complaint from the policyholder is greater than a first preset threshold within the first preset time period after underwriting is completed.

[0090] In one specific embodiment, the first preset time period can be one year, two years, or three years, etc. Taking one year as an example, in this embodiment of the application, the first complaint warning model is used to identify the probability value of a complaint occurring within one year after the new contract policy is underwritten.

[0091] It should be noted that the first complaint warning model can be built based on algorithms such as LightGBM (Light Gradient Boosting Machine) and XGBoost (Extreme Gradient Boosting) to identify the complaint risk value of new insurance policies.

[0092] In some embodiments, see Figure 7 , Figure 7 This application illustrates a policy complaint early warning method provided by an embodiment of the present application, such as... Figure 7 As shown, before a new insurance policy is underwritten, a pre-trained first complaint warning model is used to identify the new policy and send a warning result to the policy underwriting platform. This warning result includes new policies with a first complaint risk value greater than or equal to a first preset threshold. After receiving the warning result on the policy underwriting platform, relevant staff conduct a due diligence investigation on the corresponding new policy. The staff then uploads the due diligence investigation results to the policy underwriting platform. If the due diligence investigation results indicate that the new policy still has a complaint risk, the relevant staff are notified to withdraw or cancel the new policy. If the due diligence investigation results indicate that the new policy has no complaint risk or a low complaint risk, the new policy can be underwritten.

[0093] In some embodiments, see Figure 5 , Figure 5 It shows Figure 1 A flowchart illustrating the sub-steps of step S120 is shown below. Figure 5 As shown, the process of using the trained first complaint warning model to obtain the first complaint risk value based on the target policy information, the first agent information, and the first policyholder information includes, but is not limited to, steps S510 and S520.

[0094] Step S510: Extract features from the target policy information, the first agent information, and the first policyholder information to obtain the first target feature information.

[0095] Step S520: Input the first target feature information into the trained first complaint warning model, so as to output the first complaint risk value through the first complaint warning model.

[0096] It should be understandable that extracting feature information from the three dimensions of policy, agent, and policyholder enables the first complaint warning model to accurately predict the first complaint risk value of a newly signed policy. Specifically, for agent information and policyholder information, feature extraction can be performed on the basic information and behavioral information of the agent and policyholder respectively to obtain basic feature information and behavioral feature information.

[0097] In some embodiments, see Figure 2, Figure 2 This document illustrates a flowchart of the training process for the first complaint early warning model provided in an embodiment of this application. Figure 2 As shown, the first complaint warning model is trained through steps S210 to S230.

[0098] Step S210: Obtain a first historical policy sample set. The first historical policy sample set includes a first positive sample and a first negative sample. The first positive sample is a policy that has received a complaint within the first preset time period after the underwriting is completed, and the first negative sample is a policy that has not received a complaint within the first preset time period after the underwriting is completed.

[0099] Step S220: Extract features from the policy information, agent information, and policyholder information in the first positive sample and the first negative sample to obtain the first training feature information;

[0100] Step S230: With the training objective of predicting whether a complaint will occur within a first preset time period after the policy is underwritten, a preset first complaint warning model is trained based on the first training feature information to update the model parameters of the first complaint warning model until the model performance of the first complaint warning model reaches the preset first training termination condition.

[0101] In steps S210 to S230, taking the probability of identifying a complaint within one year of underwriting for a new insurance policy using the first complaint warning model as an example, a first historical policy sample set can be extracted from the policy information database, agent information database, and policyholder information database. This first historical policy sample set includes multiple policy samples, each of which has been underwritten for a full year. Policy samples with complaint records within one year of underwriting are designated as the first positive samples, and policy samples without complaint records within one year of underwriting are designated as the first negative samples. After obtaining the first historical policy sample set, feature extraction is performed on the policy information, agent information, and policyholder information in the first positive and first negative samples to obtain the first training feature information. Then, with the training objective of predicting whether a complaint will occur within one year of underwriting, the preset first complaint warning model is subjected to supervised training based on the first training feature information. The model parameters of the first complaint warning model are updated until the model performance of the first complaint warning model reaches the preset first training termination condition.

[0102] In one specific embodiment, after obtaining the first historical policy sample set, the first historical policy sample set is divided into a first training set, a first validation set, and a first cross-time validation set. For example, 80% of the policy samples underwritten from January 1, 2021 to December 31, 2021 are randomly selected as the first training set for training the model, the remaining 20% ​​of the policy samples are used as the first validation set for testing the stability of the model, and the policies underwritten from January 1, 2022 to March 1, 2022 are used as the first cross-time validation set for further validating the stability of the model over the time span.

[0103] In one specific embodiment, after each round of parameter update, the first complaint warning model is tested using the first validation set and the first cross-time validation set. If the model performance of the first complaint prediction model on the first validation set and the first cross-time validation set reaches the preset first training termination condition, then the training ends and the trained first complaint warning model is obtained.

[0104] Understandably, the performance of the First Complaint Warning Model can be evaluated using metrics such as AUC (Area Under the ROC Curve), precision, coverage, recall, and PSI (Population Stability Index).

[0105] For example, if accuracy is used as a model performance evaluation metric, the corresponding first training termination condition can be that the accuracy of the first complaint warning model is greater than or equal to 70%.

[0106] In step S140, at the first point in time after the new insurance contract is underwritten, the second agent information and the second policyholder information of the new insurance contract are obtained from the policy underwriting platform.

[0107] It should be understood that a first complaint risk value less than a first preset threshold indicates that the probability of the policyholder filing a complaint within a first preset period after the new contract policy is underwritten is low, and the policy can be underwritten.

[0108] Understandably, the target policy information, second agent information, and second policyholder information of a new contract policy can be obtained from the policy information database, agent information database, and policyholder information database pre-established in the policy underwriting platform.

[0109] It should be noted that the first time point can be the first day, the thirtieth day, the ninetieth day, etc., after the new contract policy is underwritten.

[0110] It is understandable that, immediately after a new insurance policy is underwritten, the agent's information and the policyholder's information for the new policy should be retrieved. Then, the new policy should be used to identify the risk of complaints. Establishing a model identification mechanism before and after underwriting can improve the policy complaint early warning mechanism and form a normalized system to reduce complaints.

[0111] In some embodiments, both the first agent information and the second agent information include agent behavior information. The policy complaint early warning method obtains the first agent information of the new policy through the following steps:

[0112] The first agent information for the new insurance contract is obtained from the agent information database preset in the policy underwriting platform.

[0113] It should be noted that agent behavior information includes work quality information, such as complaints received and the renewal rate of clients under the agent's name.

[0114] In some embodiments, before obtaining the second agent information and the second policyholder information of the new insurance contract from the policy underwriting platform, the method further includes:

[0115] Collect information on changes in the agent's first behavior during the period from the completion of underwriting to the first point in time for the new insurance policy;

[0116] Update the agent behavior information in the agent information database according to the first behavior change information;

[0117] The policy complaint early warning method obtains the second agent information for the new insurance policy through the following steps:

[0118] Retrieve the second agent information for the new insurance policy from the updated agent information database.

[0119] Understandably, collecting information on changes in agent behavior during the period from underwriting to the immediate point in time for new insurance policies—such as the increase in complaints against agents or the renewal rate of policies held by agents—is crucial. This information is then used to update agent behavior information in the agent information database. This allows for the retrieval of secondary agent information for new policies from the updated database at the immediate point in time, enabling a re-evaluation of policy complaint risks. Since agent behavior significantly impacts policy complaints, continuously collecting and updating agent behavior information after underwriting new policies and re-analyzing post-underwriting complaint risks from the perspective of agent behavior can improve the accuracy of complaint warning models.

[0120] In some embodiments, both the first policyholder information and the second policyholder information include policyholder behavior information. The policy complaint early warning method obtains the first policyholder information of the new policy through the following steps:

[0121] The first policyholder information of the new contract policy is obtained from the policyholder information database preset by the policy underwriting platform.

[0122] Before obtaining the second policyholder information and the second policyholder data of the new insurance contract from the policy underwriting platform, the method further includes:

[0123] Collect information on changes in the first behavior of the policyholder of the new insurance contract from the completion of underwriting to the first point in time;

[0124] Update the policyholder behavior information in the policyholder information database according to the first behavior change information;

[0125] The policy complaint early warning method obtains the second policyholder information of the new insurance contract through the following steps:

[0126] The second policyholder information for the new insurance policy is obtained from the updated policyholder information database.

[0127] Understandably, collecting information on changes in policyholder behavior between the underwriting of a new policy and the initial timeframe is crucial. This includes information such as changes in policyholder sentiment (e.g., calls or expressions of dissatisfaction on the insurance app within the last 7 / 30 / 90 days, or calls or expressions of intent to cancel the policy within the last 7 / 30 / 90 days) and financial difficulties (e.g., financial difficulties within the last 7 / 30 / 90 days). This information is then used to update the policyholder behavior database. This allows for the immediate retrieval of the second policyholder information for the new policy from the updated database, enabling a re-identification of policy complaint risks. Since policyholder behavior significantly impacts policy complaints, continuously collecting and updating policyholder behavior information after underwriting a new policy and re-analyzing post-underwriting complaint risks from the perspective of policyholder behavior can improve the accuracy of complaint early warning models.

[0128] In steps S150 and S160, the trained second complaint warning model is used to obtain a second complaint risk value based on the target policy information, the second agent information, and the second policyholder information. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point. New contract policies with a second complaint risk value greater than or equal to the second preset threshold are sent to the policy underwriting platform as high complaint risk policies.

[0129] It should be understood that a well-trained second complaint warning model uses information from three dimensions—target policy information, second agent information, and second policyholder information—as indicators to identify whether a new contract policy is a high-complaint-risk policy. This helps improve the accuracy of policy complaint warnings. A high-complaint-risk policy can be interpreted as a policy in which the probability of a complaint from the policyholder is greater than a second preset threshold within a second preset time period after the first time point.

[0130] In one specific embodiment, the first time point can be the first day, the thirtieth day, the ninetieth day, etc., after the new contract policy is underwritten. The second preset time period can be one year, two years, three years, etc.

[0131] It should be noted that if the policy complaint warning method is only used to warn whether a new contract policy will have complaints within one year, then after using the first complaint warning model to identify the complaint risk of a new contract policy before underwriting, the second complaint warning model can be used to predict whether the new contract policy will have complaints in the following period at any point in time within one year after underwriting. For example, the second complaint warning model can be used to predict the probability of a new contract policy having complaints in the following nine months in the third month after underwriting.

[0132] It should be noted that the second complaint warning model can be built based on algorithms such as LightGBM (Light Gradient Boosting Machine) and XGBoost (Extreme Gradient Boosting) to identify the complaint risk value of new insurance policies.

[0133] In some embodiments, such as Figure 7 As shown, at the first point in time, after the new contract policy is determined to be a high-complaint-risk policy through the second complaint warning model, the warning result of the new contract policy being a high-complaint-risk policy is sent to the policy underwriting platform. After receiving the warning result on the policy underwriting platform, the relevant staff will conduct a follow-up investigation on the high-complaint-risk policy. Then, the staff will upload the follow-up investigation results to the policy underwriting platform. If the follow-up investigation results indicate that the new contract policy still has a complaint risk, the relevant staff will be notified to cancel the new contract policy during the policy cooling-off period.

[0134] In some embodiments, see Figure 6 , Figure 6 It shows Figure 1 A flowchart illustrating the sub-steps of step S150 is shown below. Figure 6 As shown, the second complaint risk value is obtained by using the trained second complaint warning model based on the target policy information, the second agent information, and the second policyholder information, including but not limited to steps S610 to S620.

[0135] Step S610: Extract features from the target policy information, the second agent information, and the second policyholder information to obtain the second target feature information.

[0136] Step S620: Input the second target feature information into the trained second complaint warning model, so as to output the second complaint risk value through the second complaint warning model.

[0137] It should be understandable that extracting feature information from the three dimensions of policy, agent, and policyholder enables the second complaint early warning model to accurately predict the second complaint risk value of newly signed policies. Specifically, for agent information and policyholder information, feature extraction can be performed on the basic information and behavioral information of agent information and policyholder information respectively to obtain basic feature information and behavioral feature information.

[0138] In some embodiments, see Figure 3 , Figure 3 This document illustrates a flowchart of the training process for the second complaint early warning model provided in an embodiment of this application. Figure 3 As shown, the first complaint warning model is trained through steps S310 to S330.

[0139] Step S310: Obtain second historical policy sample data. The second historical policy sample data includes second positive samples and second negative samples. The second positive samples are policies for which complaints occurred within the second preset time period after the first time point, and the second negative samples are policies for which no complaints occurred within the second preset time period after the first time point.

[0140] Step S320: Extract features from the policy information, agent information, and policyholder information in the second positive sample and the second negative sample to obtain the second training feature information;

[0141] Step S330: With the training objective of predicting whether a complaint will arise in the second preset time period after the first time point, a preset second complaint warning model is trained based on the second training feature information to update the model parameters of the second complaint warning model until the model performance of the second complaint warning model reaches the preset second training termination condition.

[0142] In steps S310 to S330, taking the probability of identifying a new policy contract with a complaint occurring within nine months after the third month of underwriting using the second complaint warning model as an example, a second historical policy sample set can be extracted from the policy information database, agent information database, and policyholder information database. This second historical policy sample set includes multiple policy samples, each with an underwriting period of one year or more. Policy samples with complaint records within nine months after the third month of underwriting are designated as second positive samples, and policy samples without complaint records within nine months after the third month of underwriting are designated as second negative samples. After obtaining the second historical policy sample set, feature extraction is performed on the policy information, agent information, and policyholder information in the second positive and second negative samples to obtain second training feature information. Then, with the training objective of predicting whether a policy will have a complaint within nine months after the third month of underwriting, supervised training is performed on the preset second complaint warning model based on the second training feature information. The model parameters of the second complaint warning model are updated until the model performance of the second complaint warning model reaches the preset second training termination condition.

[0143] In one specific embodiment, after obtaining the second historical policy sample set, the second historical policy sample is divided into a second training set, a second validation set, and a second cross-time validation set. For example, 80% of the policy samples underwritten from January 1, 2021 to December 31, 2021 are randomly selected as the second training set for training the model, and the remaining 20% ​​of the policy samples are used as the second validation set to test the stability of the model. The policies underwritten from January 1, 2022 to March 1, 2022 are used as the second cross-time validation set to further verify the stability of the model over the time span.

[0144] In one specific embodiment, after each round of parameter update, the second complaint warning model is tested using the second validation set and the second cross-time validation set. If the model performance of the second complaint prediction model on the second validation set and the second cross-time validation set reaches the preset second training termination condition, then the training ends and the trained second complaint warning model is obtained.

[0145] Understandably, the performance of the second complaint warning model can be evaluated using metrics such as AUC (Area Under the ROC Curve), precision, coverage, recall, and PSI (Population Stability Index).

[0146] For example, if coverage is used as a model performance evaluation metric, the corresponding second training termination condition can be that the accuracy of the second complaint warning model is greater than or equal to 70%.

[0147] In some embodiments, see Figure 4 , Figure 4 This document illustrates a flowchart of the training process for the first complaint early warning model provided in an embodiment of this application. Figure 4 As shown, the method further includes steps S410 to S430.

[0148] Step S410: At the second time point after the underwriting is completed, obtain the third agent information and the third policyholder information of the new contract policy from the policy underwriting platform. The second time point is the time point after the first time point.

[0149] Step S420: Using the trained third complaint warning model, a third complaint risk value is obtained based on the target policy information, the third agent information, and the third policyholder information. The third complaint risk value represents the probability that a complaint will occur in the new contract policy within a third preset time period after the second time point.

[0150] Step S430: Send the new contract policy with the third complaint risk value greater than or equal to the third preset threshold as a high complaint risk policy to the policy underwriting platform.

[0151] Understandably, if the second complaint risk value is less than the second preset threshold, it indicates that the probability of the new contract policy being complained about by the policyholder in the second preset time period after the first time point is low. Therefore, the model identification of the new contract policy will continue at the second time point after the underwriting is completed, forming a normalized complaint reduction.

[0152] It should be noted that the specific implementation and effects of steps S410 to S430 in this embodiment can be referred to the detailed description of steps S140 to S160 above, and will not be repeated here.

[0153] In one specific embodiment, the first complaint warning model, the second complaint warning model, and the third complaint warning model are the same model, all used to predict the probability of a complaint arising within the same preset time period after the current time point in a new contract policy.

[0154] For example, a historical sample dataset is obtained, in which positive samples are policies that have received complaints within one year after underwriting, and negative samples are policies that have not received complaints within one year after underwriting. Features are extracted from the positive and negative samples, and the extracted feature information is input into the complaint warning model to conduct supervised training of the complaint warning model, resulting in a trained complaint warning model. Then, before the new contract is underwritten, on the first day, the thirtieth day, and the ninetieth day after the new contract is underwritten, the trained complaint warning model is used to identify the new contract policy and determine the probability value of whether the new contract policy will receive a complaint within one year after the current time point.

[0155] Please see Figure 8 This application also provides a policy complaint early warning device 100, which includes:

[0156] The first acquisition module 110 is used to acquire the target policy information, first agent information and first policyholder information of the new contract policy from the policy underwriting platform before the new contract policy is underwritten.

[0157] The first early warning module 120 is used to obtain a first complaint risk value based on the target policy information, the first agent information and the first policyholder information using a trained first complaint early warning model. The first complaint risk value represents the probability value of a complaint occurring within a first preset time period after the new contract policy is underwritten.

[0158] The first sending module 130 is used to send the new contract insurance policy with the first complaint risk value greater than or equal to the first preset threshold as a high complaint risk policy to the policy underwriting platform.

[0159] The second acquisition module 140 is used to acquire the second agent information and the second policyholder information of the new contract policy from the policy underwriting platform at the first time point after the new contract policy is underwritten.

[0160] The second early warning module 150 is used to obtain a second complaint risk value based on the target policy information, the second agent information and the second policyholder information using a trained second complaint early warning model. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point.

[0161] The second sending module 160 is used to send the new contract insurance policy with the second complaint risk value greater than or equal to the second preset threshold as a high complaint risk policy to the policy underwriting platform.

[0162] This application proposes a policy complaint early warning device. Before a new insurance contract is underwritten, the device obtains the target policy information, first agent information, and first policyholder information of the new contract from the policy underwriting platform. Using a first complaint risk early warning model, it obtains a first complaint risk value based on the target policy information, first agent information, and first policyholder information. The first complaint risk value represents the probability of a complaint arising within a first preset time period after the new contract is underwritten. New contracts with a first complaint risk value greater than or equal to a first preset threshold are sent to the policy underwriting platform as high-complaint-risk policies. At the first point in time after the new contract is underwritten, the device obtains the second agent information and second policyholder information of the new contract from the policy underwriting platform. Using a trained second complaint early warning model, it obtains a second complaint risk value based on the target policy, second agent information, and second policyholder information. The second complaint risk value represents the probability of a complaint arising within a second preset time period after the first point in time. New contracts with a second complaint risk value greater than or equal to a second preset threshold are then sent to the policy underwriting platform as high-complaint-risk policies. This application utilizes a complaint early warning model to predict complaints for new insurance contracts before and after policy underwriting, thereby identifying high-complaint-risk policies and achieving policy complaint early warning. On the one hand, by providing complaint early warnings for new insurance contracts, and since new policy underwriting is the source of insurance policies, improving the sales quality of new policies can fundamentally resolve and reduce policy complaints, further protecting customer rights and maintaining the company's and industry's reputation. On the other hand, by establishing a model identification mechanism before and after underwriting, the policy complaint early warning mechanism can be improved, forming a normalized complaint reduction mechanism.

[0163] It should be noted that the information interaction and execution process between the modules of the above-mentioned device are based on the same concept as the method embodiment of this application. For details on their specific functions and technical effects, please refer to the method embodiment section, and they will not be repeated here.

[0164] Please see Figure 9 , Figure 9 This application illustrates the hardware structure of an electronic device according to an embodiment of the present application. The electronic device includes:

[0165] The processor 210 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant computer programs to implement the technical solutions provided in the embodiments of this application.

[0166] The memory 220 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 220 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 220 and called and executed by the processor 210 to execute the policy complaint warning method of the embodiments of this application.

[0167] Input / output interface 230 is used to implement information input and output;

[0168] The communication interface 240 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.); and the bus 250 is used to transmit information between each component of the device (such as processor 210, memory 220, input / output interface 230 and communication interface 240).

[0169] The processor 210, memory 220, input / output interface 230 and communication interface 240 are connected to each other within the device via bus 250.

[0170] This application also provides a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more computer programs, which can be executed by one or more processors to implement the above-mentioned policy complaint early warning method.

[0171] Memory, as a computer-readable storage medium, can be used to store software programs and computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0172] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0173] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0174] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0175] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0176] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0177] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0178] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0179] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0180] Furthermore, the functional units in each embodiment of this application can be integrated into a single processing unit, or each unit can exist independently, or two or more units can be integrated into a single unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0181] If the assembled units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of each embodiment of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0182] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for early warning of policy complaints, characterized in that, The method includes: Before the new insurance contract is underwritten, obtain the target policy information, first agent information and first policyholder information of the new insurance contract from the policy underwriting platform; The first complaint risk value is obtained by using the trained first complaint warning model based on the target policy information, the first agent information and the first policyholder information. The first complaint risk value represents the probability that a complaint will occur within a first preset time period after the new contract policy is underwritten. The new insurance policy with the first complaint risk value being greater than or equal to the first preset threshold is sent as a high complaint risk policy to the policy underwriting platform. At the first point after the new insurance contract is underwritten, the information of the second agent and the second policyholder of the new insurance contract is obtained from the policy underwriting platform. The trained second complaint warning model is used to obtain a second complaint risk value based on the target policy information, the second agent information, and the second policyholder information. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point. The new insurance policy with a second complaint risk value greater than or equal to the second preset threshold is sent as a high complaint risk policy to the policy underwriting platform. Wherein, both the first policyholder information and the second policyholder information include policyholder behavior information, and the policy complaint early warning method obtains the first policyholder information of the new contract policy through the following steps: obtaining the first policyholder information of the new contract policy from the preset policyholder information database in the policy underwriting platform; Before obtaining the second agent information and the second policyholder information of the new contract policy from the policy underwriting platform, the method further includes: collecting the first behavioral change information of the policyholder of the new contract policy from the completion of underwriting to the first time point; and updating the policyholder behavioral information in the policyholder information database according to the first behavioral change information. The policy complaint early warning method obtains the second policyholder information of the new contract policy through the following steps: obtaining the second policyholder information of the new contract policy from the updated policyholder information database.

2. The policy complaint early warning method according to claim 1, characterized in that, The first complaint warning model was trained through the following steps: Obtain a first historical policy sample set, which includes a first positive sample and a first negative sample. The first positive sample is a policy that received a complaint within the first preset time period after the underwriting was completed, and the first negative sample is a policy that did not receive a complaint within the first preset time period after the underwriting was completed. Feature extraction is performed on the policy information, agent information, and policyholder information in the first positive sample and the first negative sample to obtain the first training feature information; The training objective is to predict whether a complaint will arise within a first preset time period after the policy is underwritten. Based on the first training feature information, a preset first complaint warning model is trained to update the model parameters of the first complaint warning model until the model performance of the first complaint warning model reaches the preset first training termination condition.

3. The policy complaint early warning method according to claim 2, characterized in that, The second complaint warning model is trained through the following steps: Obtain second historical policy sample data, which includes second positive samples and second negative samples. The second positive samples are policies for which complaints occurred within the second preset time period after the first time point, and the second negative samples are policies for which no complaints occurred within the second preset time period after the first time point. Feature extraction is performed on the policy information, agent information, and policyholder information in the second positive sample and the second negative sample to obtain the second training feature information; The training objective is to predict whether a complaint will arise in the second preset time period after the first time point. Based on the second training feature information, a preset second complaint warning model is trained to update the model parameters of the second complaint warning model until the model performance of the second complaint warning model reaches the preset second training termination condition.

4. The policy complaint early warning method according to claim 1, characterized in that, Both the first agent information and the second agent information include agent behavior information. The policy complaint early warning method obtains the first agent information of the new policy through the following steps: The first agent information for the new insurance contract is obtained from the agent information database preset in the policy underwriting platform. Before obtaining the second agent information and the second policyholder information of the new insurance contract from the policy underwriting platform, the method further includes: collecting information on the first behavioral changes of the agent of the new insurance contract from the completion of underwriting to the first time point; Update the agent behavior information in the agent information database according to the first behavior change information; The policy complaint early warning method obtains the second agent information for the new insurance policy through the following steps: Retrieve the second agent information for the new insurance policy from the updated agent information database.

5. The policy complaint early warning method according to claim 1, characterized in that, The method further includes: At the second point in time after the underwriting is completed, the information of the third agent and the third policyholder of the new contract policy is obtained from the policy underwriting platform. The second point in time is the time point after the first point in time. The trained third complaint warning model is used to obtain a third complaint risk value based on the target policy information, the third agent information, and the third policyholder information. The third complaint risk value represents the probability that a complaint will occur in the new contract policy within a third preset time period after the second time point. New insurance policies with a third complaint risk value greater than or equal to a third preset threshold are sent to the policy underwriting platform as high complaint risk policies.

6. The policy complaint early warning method according to claim 1, characterized in that, The step of using a trained first complaint warning model to obtain a first complaint risk value based on the target policy information, the first agent information, and the first policyholder information includes: Feature extraction is performed on the target policy information, the first agent information, and the first policyholder information to obtain the first target feature information; The first target feature information is input into the trained first complaint warning model so that the first complaint warning model can output the first complaint risk value. The step of using the trained second complaint warning model to obtain the second complaint risk value based on the target policy information, the second agent information, and the second policyholder information includes: extracting features from the target policy information, the second agent information, and the second policyholder information to obtain second target feature information; The second target feature information is input into the trained second complaint warning model, so that the second complaint risk value is output through the second complaint warning model.

7. A policy complaint early warning device, characterized in that, The device includes: The first acquisition module is used to acquire the target policy information, first agent information and first policyholder information of the new contract policy from the policy underwriting platform before the new contract policy is underwritten. The first early warning module is used to obtain a first complaint risk value based on the target policy information, the first agent information, and the first policyholder information using a trained first complaint early warning model. The first complaint risk value represents the probability that a complaint will occur within a first preset time period after the new contract policy is underwritten. The first sending module is used to send the new insurance contract with the first complaint risk value greater than or equal to the first preset threshold as a high complaint risk policy to the policy underwriting platform. The second acquisition module is used to acquire the second agent information and the second policyholder information of the new contract policy from the policy underwriting platform at the first time point after the new contract policy is underwritten. The second early warning module is used to obtain a second complaint risk value based on the target policy information, the second agent information, and the second policyholder information using a trained second complaint early warning model. The second complaint risk value represents the probability that a complaint will occur in the new contract policy within a second preset time period after the first time point. The second sending module is used to send the new insurance policy with the second complaint risk value greater than or equal to the second preset threshold as a high complaint risk policy to the policy underwriting platform. Wherein, both the first policyholder information and the second policyholder information include policyholder behavior information, and the policy complaint early warning device is also used to obtain the first policyholder information of the new contract policy through the following steps: obtaining the first policyholder information of the new contract policy from the policyholder information database preset in the policy underwriting platform; Before obtaining the second agent information and the second policyholder information of the new contract policy from the policy underwriting platform, the policy complaint early warning device is also used to: collect the first behavior change information of the policyholder of the new contract policy from the completion of underwriting to the first time point; and update the policyholder behavior information in the policyholder information database according to the first behavior change information. The policy complaint early warning device is also used to obtain the second policyholder information of the new contract policy through the following steps: obtaining the second policyholder information of the new contract policy from the updated policyholder information database.

8. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that is executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images