Risk identification method, device and equipment

Abstract

The embodiment of the invention provides a risk identification method, device and equipment. The scheme comprises the following steps: acquiring insurance buying information; wherein the insurance buying information at least comprises insurance disclosure information; obtaining a pre-stored risk identification prompt word for the insurance buying information; wherein the risk identification cue word comprises rule description information for a risk identification rule, and first thinking chain information and second thinking chain information which reflect risk identification logic of the risk identification rule, and the first thinking chain information is used for describing information for identifying a risk factor from insurance buying information to obtain a factor identification result; the second thinking chain information is used for describing information of a rule recognition result obtained through logic judgment based on the factor recognition result; and inputting the risk identification cue word and the insurance buying information into a large language model to obtain a risk identification result for the insurance buying information determined by the large language model.

Description

Technical Field

[0001] The present application relates to the field of risk management technology, and in particular to a risk identification method, a risk identification device, and a risk identification equipment. Background Art

[0002] The sales process of insurance products usually involves various agreement documents such as the insurance application page, product terms, insurance instructions, disclaimers, special agreements, and electronic insurance policies. These agreement documents may contain risks such as non-standard and unreasonable expressions, and inconsistent content. These risks often affect user rights or claims matters. Therefore, it is necessary to identify the risks in insurance products.

[0003] Existing technologies primarily rely on pre-defined rules to identify whether an insurance product presents risks. For example, this approach involves identifying keywords within agreement documents. However, these rules are tailored to specific inherent risks. If new risks emerge after an insurance product is updated or a new product is released, the existing rules will no longer apply. This demonstrates the limited flexibility and generalizability of existing risk identification methods for insurance product risk identification.

[0004] Therefore, how to improve the flexibility and generalization ability of risk identification for insurance products has become a technical problem that needs to be solved urgently. Summary of the Invention

[0005] In view of this, embodiments of the present application provide a risk identification method, apparatus, and device to address the problems of poor flexibility and generalization capability in risk identification for insurance products.

[0006] According to a first aspect of an embodiment of the present application, a risk identification method is provided, which is applied to a risk identification server, including: obtaining insurance application information; the insurance application information includes at least insurance disclosure information; obtaining pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for risk identification rules and thought chain information reflecting the risk identification logic of the risk identification rules; the thought chain information includes first thought chain information describing the factor identification result obtained by identifying risk factors from the insurance application information, and second thought chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result; the risk identification prompt words and the insurance application information are input into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

[0007] According to a second aspect of an embodiment of the present application, a risk identification device is provided, including: an insurance application information acquisition module, used to obtain insurance application information; the insurance application information includes at least insurance disclosure information; a prompt word acquisition module, used to obtain pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for risk identification rules and thinking chain information reflecting the risk identification logic of the risk identification rules; the thinking chain information includes first thinking chain information describing the identification of risk factors from the insurance application information to obtain a factor identification result, and second thinking chain information describing the logical judgment based on the factor identification result to obtain a rule identification result; a risk identification module, used to input the risk identification prompt words and the insurance application information into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

[0008] According to a third aspect of an embodiment of the present application, a risk identification device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: obtain insurance application information; the insurance application information includes at least insurance disclosure information; obtain pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for risk identification rules and thinking chain information reflecting the risk identification logic of the risk identification rules; the thinking chain information includes first thinking chain information describing the factor identification result obtained by identifying the risk factor from the insurance application information, and second thinking chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result; the risk identification prompt words and the insurance application information are input into a large language model to obtain the risk identification result for the insurance application information determined by the large language model.

[0009] At least one embodiment of the present specification can achieve the following beneficial effects: pre-generating risk identification prompts for risk items in insurance application information, inputting the risk identification prompts and the insurance application information into a large language model, and determining the risk identification results of the insurance application information by the large language model. The risk identification prompts include rule description information for a risk identification rule and first and second thought chain information reflecting the risk identification logic of the risk identification rule. The first thought chain information describes the factor identification result obtained by identifying risk factors from the insurance application information, and the second thought chain information describes the rule identification result obtained by performing logical judgment based on the factor identification result. In this solution, the risk identification prompts input into the large language model include rule description information for the risk identification rule and thought chain information reflecting the risk identification logic of the risk identification rule, so that the large language model can understand the risk identification intent based on the rule description information of the risk identification rule, and can perform risk identification on the insurance application information according to the thought chain information reflecting the risk identification logic of the risk identification rule, thereby utilizing the language understanding ability and high scalability of the large language model to identify risks present in the insurance application information, thereby improving the flexibility and generalization of risk identification for insurance products. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0011] Figure 1 This is a schematic diagram of an application scenario of a risk identification method provided in an embodiment of this specification;

[0012] Figure 2 This is a flow chart of a risk identification method provided in an embodiment of this specification;

[0013] Figure 3 This is a schematic diagram of a configuration page for risk identification rules provided in an embodiment of this specification;

[0014] Figure 4 This is a swim lane flow diagram of a risk identification method provided in an embodiment of this specification;

[0015] Figure 5 This specification provides the corresponding embodiment Figure 2 A schematic structural diagram of a risk identification device;

[0016] Figure 6This specification provides the corresponding embodiment Figure 2 A structural block diagram of a risk identification device. DETAILED DESCRIPTION

[0017] The following description sets forth many specific details to facilitate a thorough understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of the present application. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0018] The terms used in one or more embodiments of the present application are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of the present application. The singular forms "a", "the" and "the" used in one or more embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of the present application refers to and includes any or all possible combinations of one or more associated listed items.

[0019] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0020] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0021] First, the terms involved in one or more embodiments of the present application are explained.

[0022] Chain of Thought (COT) is a method of generating more accurate and coherent output by guiding the model to reason step by step. The traditional prompt design usually directly gives a question or task, and the model generates output information based on the input. However, this method often ignores the intermediate steps of the model in the reasoning process, resulting in the output information may not be accurate or logical. The core idea of the Chain of Thought COT is to guide the model to think according to a logical chain by introducing a series of step-by-step reasoning steps in the Prompt. In this way, the model can not only generate the final answer, but also show the reasoning process, thereby improving the accuracy and interpretability of the output. In the embodiments of this specification, the prompt words used may contain information reflecting the chain of thought, that is, the chain of thought prompt information.

[0023] An agent, or AI agent model, is a highly intelligent model, distinct from traditional machine learning models. It can perceive its environment, make decisions, and execute tasks. The core of the AI agent model lies in its ability to think independently and utilize tools, enabling it to gradually achieve a given goal. During the use of the agent model, it can gradually complete a specified task based on predefined roles and path planning.

[0024] Since each insurance product typically involves hundreds of pages of insurance documents, and the number of policies gradually increases as policies are issued after the product is launched, and the insurance documents are extremely professional and complex, manual review of insurance documents is ineffective. Risk identification based on risk identification rules requires exhaustive enumeration of risky keyword information, which incurs significant maintenance costs. Furthermore, when new insurance products emerge, existing risk identification rules are unable to cover the new risk items they introduce, resulting in limited flexibility and generalizability in risk identification for specific insurance products.

[0025] The technical solutions provided by the embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0026] Figure 1 This is a schematic diagram of an application scenario of a risk identification method provided in an embodiment of this specification.

[0027] like Figure 1 As shown, the application scenario diagram includes server 101 and server 102.

[0028] In the embodiments of this specification, the server may include but is not limited to any device, equipment, platform, equipment cluster or cloud computing service center with computing and processing capabilities. Server 101 and server 102 may be any of the aforementioned servers.

[0029] A communication connection may be established directly or indirectly between server 101 and server 102, wherein the communication connection may include, but is not limited to, a local area network connection, a wide area network connection, an Internet connection, a short-range communication connection, or other types of data network connections. Short-range communication connections include, but are not limited to, near field communication (NFC), local area network, Bluetooth, infrared, and other connection methods.

[0030] In actual application, when an insurance product is launched on the platform, or a new insurance policy is generated for an insurance product on the platform, or when the platform conducts an inspection on an insurance product that has been launched, risk identification for the insurance product can be triggered. Server 101 can be a server that triggers server 102 to perform risk identification, and server 102 can be a server that performs a risk identification operation based on the triggering operation of server 101. After the risk identification for the insurance product is triggered, server 101 can send insurance application information to server 102. Server 102 calls the intelligent agent to perform insurance risk identification on the insurance application information based on the received insurance application information. Server 102 sends the risk identification result returned by the intelligent agent after risk identification of the insurance application information to server 101, so that the user can display the risk identification result for the insurance application information through the user terminal corresponding to server 101.

[0031] In this application, a risk identification method is provided, which is described in detail one by one in the following embodiments.

[0032] Figure 2 This is a flow chart of a risk identification method provided in an embodiment of this specification.

[0033] From a program perspective, the execution body of the process can be a program installed on the risk identification server. It is understood that the method can be executed by any device, equipment, platform, or device cluster with computing and processing capabilities.

[0034] like Figure 2 As shown, the process may include the following steps.

[0035] Step 202: Obtain insurance application information; the insurance application information at least includes insurance disclosure information.

[0036] Optionally, obtaining insurance application information may include obtaining insurance application information of the insurance product to be identified.

[0037] In the embodiments of this specification, the insurance application information may include insurance disclosure information, wherein the insurance disclosure information may include information such as the application page, application instructions, insurance terms, disclaimers, health notices, special agreements, etc. The application page may also be referred to as the application interface or the application page, which usually refers to the electronic page where the user fills in the application information, confirms the insurance terms and completes the payment online when purchasing an insurance product. Application instructions may refer to the basic explanatory document that must be read before purchasing insurance. Insurance terms may refer to the legal text of the insurance contract, which is used to clarify the coverage responsibilities and the rights and obligations of both parties. Disclaimers may refer to a list of special circumstances in which the insurance company does not bear liability for compensation. Health notices may refer to a written statement of the user's health status when taking out insurance. Special agreements may refer to personalized supplements or modifications to standard terms, which have the same legal effect as standard terms.

[0038] Optionally, insurance application information may also include insurance disclosure information and policy information. Policy information is the written proof of the insurance contract signed between the insurer and the policyholder. Policy information, also known as insurance policy information, is proof of the establishment of an insurance contract. It fully records the rights, obligations and responsibilities of both parties to the insurance contract and is the basis for the performance of the contract by both parties. Typically, policy information may include detailed information such as policyholder information, insured information, premium information, insurance coverage, terms, scope of liability, insurance amount, etc. The policy information may be desensitized policy information. Desensitization can be a key means to balance data utilization and privacy protection. Desensitization of policy information can support insurance business operations while minimizing the risk of user information leakage. In practical applications, desensitization can be performed by blurring the user information in insurance information. For example, the original user information in the insurance policy information may be: Zhang San, policy number 123456789123456, ID card 123456789123456789, mobile phone 12345678900. After desensitization, the user information in the insurance policy information may be: Zhang*, policy number 1234*******3456, ID card 123456*********789, mobile phone 123****8900.

[0039] In the embodiments of this specification, the insurance information may be information obtained from a product data platform, and the product data platform may be any platform that can provide information about insurance products.

[0040] From the perspective of data sources, in actual business, the product data platform may include platforms such as product center, sales center, liability center, health reporting center, policy center, and legal center. Insurance information can be data directly obtained from each decentralized product data platform; or, it can be data pre-collected and integrated from various decentralized product data platforms obtained from a comprehensive product data platform.

[0041] From the perspective of the method execution process, in actual business, the insurance information can be data directly obtained by the risk identification server from the product data platform, or it can be data obtained from the product data platform via the rule engine and then transmitted to the risk identification server.

[0042] Step 204: Obtain pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for the risk identification rules and thought chain information reflecting the risk identification logic of the risk identification rules; the thought chain information includes first thought chain information describing the factor identification result obtained by identifying the risk factors from the insurance application information, and second thought chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result.

[0043] Optionally, the obtaining of pre-stored risk identification prompt words for the insurance application information may include obtaining pre-stored risk identification prompt words for the insurance product to be identified.

[0044] In the embodiments of this specification, the risk identification prompt word may be a prompt word information pre-generated based on the risk identification rules applicable to the insurance product to be identified. One insurance product may correspond to N risk identification rules (N is a positive integer), and each risk identification rule corresponds to one risk identification prompt word information.

[0045] In the embodiments of this specification, the timing for risk identification for insurance products may include the insurance product release stage when the insurance company releases the insurance product, or the maintenance stage when routine maintenance and inspection of the insurance product is carried out after the insurance product goes online.

[0046] In order to facilitate those skilled in the art to understand this solution, this specification provides a specific example 1 for risk identification prompt words.

[0047] In example 1, the risk identification prompt words may include ##rule description and ##COT, where ##COT may include step 1, step 2, and step 3.

[0048] ##Rule description: When the medical and critical illness insurance terms / insurance instructions contain descriptions of restricted occupations, other agreements need to include a list of occupations or describe occupational restrictions (see terms and insurance instructions).

[0049] ##COT:

[0050] Step 1: Please check the terms and conditions and insurance information to see if there are any occupational restrictions for the insured. If there are no restrictions, this rule will not be triggered.

[0051] If the terms or insurance information mentions that there are occupational restrictions on the insured, continue with the judgment.

[0052] Step 2: Determine the following factors:

[0053] 1. Please go to the disclaimer to check whether there is a list of occupations / which specific occupations are insurable / which specific occupations are not insurable.

[0054] 2. Please go to the Terms and Conditions / Insurance Notices to check whether there is a list of occupations / which specific occupations are insurable / which specific occupations are not insurable.

[0055] 3. Please go to the health notice to find out whether there is a list of occupations mentioned / which specific occupations can be insured / which specific occupations cannot be insured.

[0056] Step 3: If the above 1, 2, and 3 conditions are met, the rule output is: no risk;

[0057] If any of the above 1, 2, and 3 is not met, the output is: there is a risk, what materials are missing, what statements are missing, or what is missing in the expression method.

[0058] In Example 1, the rule description information may refer to the content corresponding to ##rule description, and the thinking chain information may refer to the content corresponding to ##COT.

[0059] Risk factors may include a first risk factor and a second risk factor, wherein the first risk factor may include at least one of whether the terms and conditions mention any occupational restrictions on the insured, and whether the insurance instructions mention any occupational restrictions on the insured. The second risk factor may include at least one of whether the disclaimer mentions a list of occupations / which specific occupations are insurable / which specific occupations are not insurable, whether the terms and conditions mention a list of occupations / which specific occupations are insurable / which specific occupations are not insurable, whether the insurance instructions mention a list of occupations / which specific occupations are insurable / which specific occupations are not insurable, and whether the health notice mentions a list of occupations / which specific occupations are insurable / which specific occupations are not insurable.

[0060] The type of the factor identification result for any of the first risk factors may include “yes” or “no”; the type of the factor identification result for any of the second risk factors may include “yes” or “no”.

[0061] The logical judgment may include a first logical judgment on the factor identification result of the first risk factor and a second logical judgment on the factor identification result of the second risk factor. For example, in Example 1, if the factor identification results for the first risk factor are all "no," the first logical judgment will not trigger subsequent operations, that is, the factor identification operation for the second risk factor will not be executed. If the factor identification result for the first risk factor is "yes," the first logical judgment will trigger subsequent operations, that is, the factor identification operation for the second risk factor will be executed.

[0062] For example, in Example 1, if the factor identification results for the second risk factor are all "yes", then after the second logical judgment, the output rule identification result is "no risk"; if there is a "no" in the factor identification results for the second risk factor, then after the second logical judgment, the output rule identification result is "risky", and the specific output is "what materials, what statements are missing, or what is missing in the expression method". Assume that the factor identification result for the second risk factor "Does the disclaimer mention the list of occupations / which specific occupations can be insured / which specific occupations cannot be insured" is "no", and the factor identification results for other second risk factors are "yes", then after the second logical judgment, the output rule identification result is "risky, the disclaimer does not mention the list of occupations / which specific occupations can be insured / which specific occupations cannot be insured."

[0063] Step 206: Input the risk identification prompt word and the insurance application information into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

[0064] In the embodiments of this specification, the risk identification result may include a rule identification result obtained after logical judgment on the factor identification result of the risk factor, or the risk identification result may also include a factor identification result and a rule identification result, which is not specifically limited.

[0065] It should be understood that the order of some steps in the methods described in one or more embodiments of this specification can be interchanged according to actual needs, or some steps can be omitted or deleted.

[0066] Figure 2In the method, the risk identification prompt words input into the large language model include rule description information of the risk identification rules and thought chain information reflecting the risk identification logic of the risk identification rules, so that the large language model can obtain the risk identification intention based on the rule description information of the risk identification rules, and can perform risk identification on the insurance application information according to the thought chain information reflecting the risk identification logic of the risk identification rules, so as to utilize the language understanding ability and high scalability of the large language model to identify the risks existing in the insurance application information, so as to improve the flexibility and generalization ability of risk identification for insurance products.

[0067] based on Figure 2 The present specification also provides some improved implementation methods of the method, which are described below.

[0068] In order to improve the convenience of obtaining risk identification prompt words, when storing risk identification prompt words, the corresponding risk identification prompt words may be stored based on the rule identifier of the risk identification rule.

[0069] Optionally, before obtaining the pre-stored risk identification prompt word for the insurance application information, the method may further include: receiving a risk identification request sent by a risk management server; the risk identification request carries a rule identifier of a risk identification rule applicable to risk identification of the insurance application information;

[0070] Correspondingly, the obtaining of pre-stored risk identification prompt words for the insurance application information may specifically include: obtaining pre-stored risk identification prompt words corresponding to the rule identifier; the risk identification prompt words are obtained by rewriting the risk identification rules corresponding to the rule identifier.

[0071] In the embodiments of this specification, the risk management server and the risk identification server can be the same server, or they can be independent servers. The risk management server and the risk identification server can be deployed on the same platform for risk identification of insurance products, or they can be deployed on different platforms for risk identification of insurance products, without limitation.

[0072] In an embodiment of the present specification, the risk identification request may be an identification request generated by the risk management server for the insurance product to be identified. The timing of generating the risk identification request may include generating the identification request when the insurance product to be identified goes online, or periodically generating the identification request for the insurance product to be identified based on daily inspection needs.

[0073] The risk identification request may include a rule identifier for a risk identification rule applicable to risk identification of insurance application information. The risk identification rule may be generated by a rule engine based on the insurance application information. The rule engine may be deployed within a risk management server or a server other than the risk management server, without limitation. The rule identifier may be an identifier used to uniquely identify the risk identification rule. In practical applications, the rule identifier may include an identifier used to reflect the risk type and an identifier used to reflect rule ranking information.

[0074] The risk identification request may also include insurance information of the insurance product to be identified.

[0075] In an embodiment of this specification, the risk identification server can pre-rewrite the risk identification rules to obtain risk identification prompt words, and store the generated risk identification prompt words in the risk identification server, as well as the association relationship between the risk identification prompt words and the rule identifier of the risk identification rules.

[0076] In an embodiment of the present specification, after receiving a risk identification request from a risk management server, a risk identification server may parse the risk identification request to obtain insurance application information of the insurance product to be identified, and may also simultaneously obtain a rule identifier for a risk identification rule corresponding to the insurance application information from the risk identification request. The risk identification server may use the obtained rule identifier and, based on an association between the rule identifier and a risk identification prompt word, retrieve the risk identification prompt word corresponding to the rule identifier from the risk identification server. The server may then simultaneously input the obtained insurance application information and the risk identification prompt word into a large language model, thereby enabling the large language model to perform risk identification on the insurance application information based on the risk identification prompt word.

[0077] In the embodiment of this specification, the risk identification server can query the risk identification prompt word corresponding to the rule identifier based on the association relationship between the rule identifier and the risk identification prompt word, thereby improving the convenience and accuracy of obtaining the risk identification prompt word.

[0078] Different types of insurance products have corresponding risk identification rules that are different. Therefore, in order to improve the effectiveness of the risk identification rules used when identifying risks of insurance products based on risk identification rules, the risk identification rules corresponding to the insurance products can be determined based on the type information of the insurance products.

[0079] Optionally, before receiving the risk identification request sent by the risk management server, it may also include: the risk management server obtains the insurance type information of the insurance product corresponding to the insurance application information; and determines the risk identification rules applicable to risk identification of the insurance application information based on the insurance type information.

[0080] In the embodiment of this specification, the risk management server can obtain insurance type information of insurance products from insurance application information. Based on the insurance type information, insurance products may include accident insurance, medical insurance, life insurance, travel insurance, critical illness insurance, property insurance, health insurance, etc.

[0081] In practical applications, a rule engine deployed in a risk management server can filter risk identification rules corresponding to the risk type information from a risk identification rule library based on the risk type information of the insurance product to be identified. The risk identification rule library can be a database consisting of pre-generated risk identification rules based on the risk type information of various insurance products. Different risk identification rules can be applicable to different types of insurance products. For example, some risk identification rules can be applicable to all types of insurance products, some can be applicable to travel insurance but not life insurance, and some can be applicable to life insurance but not travel insurance, etc.

[0082] In the embodiments of this specification, risk identification rules used for risk identification of insurance products are determined based on the risk type information of insurance products, thereby improving the effectiveness of the risk identification rules used and further improving the accuracy of risk identification of insurance products.

[0083] In order to save risk identification resources, risk identification operations can be triggered for insurance products when the insurance products meet preset conditions.

[0084] Optionally, before the risk management server obtains the insurance type information of the insurance product corresponding to the insurance application information, it may also include: the risk management server receives a risk identification trigger request sent by the business server; the risk identification trigger request includes at least one of a first risk identification trigger request for triggering the identification of product risks of the insurance product to be released and a second risk identification trigger request for triggering the identification of product risks of the insurance product that has been released.

[0085] In the embodiments of this specification, the business server, risk management server, and risk identification server may be the same server, or they may all be independent servers. The business server, risk management server, and risk identification server may all be deployed on the same platform for risk identification of insurance products, or they may be deployed on separate platforms for risk identification of insurance products, without limitation.

[0086] In the embodiments of this specification, a risk identification trigger request is used to trigger a risk management server to generate a risk identification request. The types of risk identification trigger requests may include at least one of a first risk identification trigger request for triggering the identification of product risks of an insurance product to be released, and a second risk identification trigger request for triggering the identification of product risks of an insurance product that has been released. In actual applications, if the insurance product to be identified has not been released, then when it is monitored that the insurance product to be identified has been released, the business server may generate a first risk identification trigger request for the insurance product to be identified; if the insurance product to be identified has been released, then when the set conditions are met, the business server may generate a second risk identification trigger request for the insurance product to be identified, wherein the set conditions may be to periodically generate a risk identification trigger request for the insurance product to be identified according to a preset time period.

[0087] In the embodiments of this specification, a risk identification trigger request is generated for an insurance product only when the insurance product meets preset conditions, thereby saving identification resources of the risk identification platform and improving the utilization rate of identification resources.

[0088] Optionally, after inputting the risk identification prompt words and the insurance application information into the large language model and obtaining the risk identification result for the insurance application information determined by the large language model, it may also include: sending the risk identification result to the rule engine in the risk management server; so that the risk identification result is sent to the business server via the rule engine, and then displayed to the user by the user terminal corresponding to the business server.

[0089] In order to improve the efficiency of risk identification for insurance products, risk identification prompt words can be pre-generated based on risk identification rules.

[0090] Optionally, before obtaining the pre-stored risk identification prompt words for the insurance application information, the method may also include: obtaining risk identification rules for identifying target risk information in the insurance application information; and rewriting the risk identification rules to obtain risk identification prompt words.

[0091] In the embodiments of this specification, the target risk information may be the risk points that need attention that are statistically summarized by risk experts for risk products based on information such as laws and regulations, compliance requirements, and historical public opinion. The target risk information may usually be expressed in natural language, and the risk identification rules may be rules generated based on the target risk information.

[0092] In the embodiments of this specification, risk identification rules can be rewritten into risk identification prompt words using pre-set program code, where each risk identification rule can correspond to a risk identification prompt word. During the rewriting of risk identification rules, each risk identification rule only needs to be rewritten once, and the rewritten risk identification prompt words are stored in the risk identification server. The risk identification server can subsequently call upon the stored risk identification prompt words as needed.

[0093] In the embodiments of this specification, the risk identification rules of insurance products are pre-written as risk identification prompt words and saved in the risk identification server. When the risk identification server subsequently performs risk identification on the insurance product, the risk identification server can directly call the risk identification prompt words for risk identification to improve the efficiency of risk identification for the insurance product.

[0094] In the embodiments of this specification, a specific embodiment is also proposed for a method of generating risk identification rules for identifying target risk information.

[0095] Optionally, the acquisition of risk identification rules for identifying target risk information in the insurance application information may specifically include: receiving a user's rule configuration operation for target risk information; the rule configuration operation includes a first configuration operation for configuring the risk factor to be identified, and a second configuration operation for configuring the logical judgment process executed for the risk factor; in response to the rule configuration operation, generating a risk identification rule for identifying the target risk information.

[0096] In the embodiments of this specification, the user may be a person responsible for risk identification for insurance products. A rule configuration operation may be an operation for the user to generate risk identification rules based on key information extracted from target risk information. The rule configuration operation may include a first configuration operation and a second configuration operation. The first configuration operation may be an operation for configuring the generation of a risk factor to be identified based on the extracted key information, and the second configuration operation may be an operation for configuring the logical judgment process executed for the risk factor.

[0097] In the embodiments of this specification, the target risk information may be content to be identified expressed in natural language. For example, when the terms and conditions / insurance instructions contain descriptions of restricted occupations, other agreements may require a list of occupations or a description of occupational restrictions.

[0098] In the embodiments of this specification, the rule engine may generate risk identification rules for identifying target risk information in response to a user's rule configuration operation.

[0099] In order to facilitate those skilled in the art to understand the present solution, this specification provides a specific example for risk identification rules.

[0100] Figure 3 This is a schematic diagram of a configuration page for risk identification rules provided in an embodiment of this specification. Figure 3 This includes the risk factors generated after the first configuration operation for "*Material Factors Involved", "Whether to mention occupational restrictions on the insured" and "Whether to mention the occupational list / which specific occupations can be insured / which specific occupations cannot be insured". Figure 3 It also includes the logical judgment expression "|(F<1>, F<2>)" generated after the second configuration operation is performed on the "*trigger condition expression", and the logical judgment expression "&(|(F<3>, F<8>), |(F<4>, F<6>), |(F<5>, F<12>))" generated after the second configuration operation is performed on the "*detection condition expression:". Figure 3 It can also include rule identification information of risk identification rules, target risk information used to generate risk identification rules ("When there are restricted occupational insurance descriptions in the terms / insurance instructions, other agreements need to include an occupational list or description of occupational restrictions"), preset problem description information, and information used to indicate whether there is a risk issue.

[0101] In the embodiments of this specification, another specific embodiment is proposed for a method of generating risk identification rules for identifying target risk information.

[0102] Optionally, the acquisition of risk identification rules for identifying target risk information in the insurance application information may specifically include: acquiring target risk information provided by a user; acquiring rule generation prompt words; the rule generation prompt words including rule generation task information, rule structure information and rule case information; the rule generation task information is used to instruct the first intelligent agent to refer to the rule case information to rewrite the target risk information into a risk identification rule that conforms to the structure represented by the rule structure information; inputting the target risk information and the rule generation prompt words into the first intelligent agent to obtain the risk identification rules output by the first intelligent agent.

[0103] In an embodiment of the present specification, a corresponding rule generation prompt word is generated based on the target risk information provided by the user, and the target risk information can be rewritten into a risk identification rule based on the rule generation prompt word. Among them, the rule generation prompt word may include rule generation task information, rule structure information and rule case information. The rule structure information may be information for limiting the structure of the risk identification rule, the rule case information may include rule positive case information and rule negative case information, and the rule task information may be information for instructing the first intelligent agent to refer to the rule case information to rewrite the target risk information into a risk identification rule that conforms to the structure represented by the rule structure information. The rule generation prompt word may also include thinking chain information for instructing the first intelligent agent to generate a risk identification rule according to certain logical thinking steps.

[0104] In the embodiments of this specification, agents and target risk information may correspond to each other. One target risk information may correspond to one agent, and one agent may correspond to multiple target risk information. After obtaining the target risk information and rule-generated prompts, the target risk information and rule-generated prompts may be input to a corresponding first agent. The first agent may then utilize a corresponding large language model to rewrite the target risk information into a risk identification rule based on the rule-generated prompts.

[0105] In the embodiments of this specification, target risk information can be automatically rewritten into risk identification rules based on a large language model, thereby improving the efficiency of rewriting risk identification rules and saving human resource costs.

[0106] Optionally, inputting the risk identification prompt words and the insurance application information into a large language model to obtain a risk identification result for the insurance product determined by the large language model may specifically include: inputting the risk identification prompt words and the insurance application information into a second intelligent agent to obtain a risk identification result for the insurance product determined by the second intelligent agent by calling the large language model.

[0107] In order to improve the efficiency of risk identification for insurance products, large language models can simultaneously identify risk items in multiple aspects of insurance products.

[0108] Optionally, the risk identification prompt words specifically include N risk identification prompt words; N is a positive integer; inputting the risk identification prompt words and the insurance application information into a large language model to obtain a risk identification result for the insurance application information determined by the large language model may specifically include: assembling the N risk identification prompt words into assembled prompt words for input into a second intelligent agent; the assembled prompt words contain the N risk identification prompt words and intelligent agent task information; the intelligent agent task information is used to instruct the second intelligent agent to determine N single risk identification results corresponding to the N risk identification prompt words, and determine an overall risk identification result based on the N single risk identification results; one single risk identification result corresponds to one risk identification prompt word, and different single risk identification results correspond to different risk identification prompt words; inputting the assembled prompt words and the insurance application information into the second intelligent agent to obtain a risk identification result for the insurance product determined by the second intelligent agent by calling the large language model.

[0109] In the embodiments of this specification, the insurance application information for an insurance product may involve multiple aspects of insurance risk. That is, the insurance application information may include multiple target risk information. Each target risk information can be rewritten as a risk identification rule, and each risk identification rule can also be converted into a risk identification prompt word. Therefore, N risk identification prompt words can be generated for the insurance application information of the insurance product.

[0110] In an embodiment of the present specification, N risk identification prompt words are combined to obtain an assembled prompt word corresponding to the insurance application information. The assembled prompt word may include N risk identification prompt words and agent task information, and the agent task information is used to instruct the second agent to generate N single risk identification results for the N risk identification prompt words, and to instruct to determine the overall risk identification result based on the N single risk identification results. The single risk identification result is the identification result generated for one risk identification prompt word, and the overall risk identification result is the identification result generated for the assembled prompt word. Among them, a single risk identification result corresponds to one of the risk identification prompt words, and different single risk identification results correspond to different risk identification prompt words.

[0111] In an embodiment of this specification, the assembly prompt words and insurance information are input into the second intelligent agent. The second intelligent agent can call the large language model based on the assembly prompt words to identify the risks of the insurance information and obtain the risk identification results corresponding to the insurance product.

[0112] In actual applications, the risk identification result may include the overall risk identification result, or the risk identification result may also include some single risk identification results. For example, the risk identification result may only include non-compliant risk identification results among N single risk identification results, but not compliant risk identification results.

[0113] In the embodiment of this specification, the assembly prompt words may also include the role information of the second intelligent agent, the thinking chain information for instructing the second intelligent agent to generate risk identification results according to certain logical thinking steps, the output format information of the risk identification results, etc.

[0114] In order to facilitate those skilled in the art to understand the assembly prompt information, this specification provides a specific example 2 for the assembly prompt information.

[0115] Example 2: Assembling prompt words.

[0116] Risk Identification Tips 1

[0117] Risk Identification Tips 2 ......

[0119] Risk Identification Tips N-1

[0120] Risk Identification Tips N

[0121] ###Role: Insurance product compliance risk expert.

[0122] ###Task: Identify compliance-related risks across all insurance product terms, insurance types, electronic policies, liabilities, graphic descriptions, health notices, coverage details, exemptions, insurance policy notes, special agreements, etc. You will need to identify all risk rules that the system needs to monitor, verify each one, and provide verification results.

[0123] ###Step 1: Determine the scope of rules applicable to the current product. For example, medical insurance only needs to run product rules applicable to health insurance;

[0124] Step 2: Analyze the rules one by one to determine the trigger conditions and find out what materials are needed to determine whether risk trigger factors are involved;

[0125] Step 3: If the corresponding material triggers risk rule verification, go to the corresponding risk verification material to check whether it contains the verification factor;

[0126] Step 4: Determine whether there is risk based on the query results and verification logic of the factors;

[0127] Step 5: Output the parsing results of each rule;

[0128] Step 6: Output the overall analysis results, risk level, and error cause;

[0129] Step 7: Check carefully to avoid omissions, duplications, and misjudgments. For any information that cannot be confirmed or is missing, please return the warning.

[0130] Output format

[0131] {Is the product risky: Yes, No

[0132] Name of the rule that triggers the risk

[0133] Triggering Cause

[0134] Verification failure reason

[0135] }

[0136] In an embodiment of the present specification, multiple risk identification prompt words can be assembled into assembly prompt words, and the assembly prompt words can be input into the second intelligent agent so that the second intelligent agent can call a large language model to simultaneously perform risk identification on multiple risk angles of insurance products based on multiple risk identification prompt words, so as to improve the efficiency of risk identification for insurance products.

[0137] In order to improve the accuracy of risk identification, for each risk identification task, the intelligent agent can call a large language model that can handle the risk identification task to process the risk identification task.

[0138] Optionally, the step of inputting the assembly prompt words and the insurance application information into the second intelligent agent to obtain a risk identification result for the insurance product determined by the second intelligent agent by calling a large language model may specifically include: inputting the assembly prompt words and the insurance application information into the second intelligent agent; calling a first large language model by the second intelligent agent to perform a first recognition task corresponding to a first prompt word among the N risk identification prompt words to obtain a first single risk identification result; calling a second large language model by the second intelligent agent to perform a second recognition task corresponding to a second prompt word among the N risk identification prompt words to obtain a second single risk identification result; obtaining a risk identification result for the insurance product determined by the second intelligent agent based on the N single risk identification results; the N single risk identification results include the first single risk identification result and the second single risk identification result.

[0139] In an embodiment of the present specification, after the assembly prompt words are input into the second intelligent agent, the second intelligent agent can identify each risk identification prompt word from the assembly prompt words. The second intelligent agent determines the large language model corresponding to the risk identification prompt word based on the correspondence between the risk identification prompt word and the large language model, and then calls the large speech model to perform the recognition task corresponding to the risk identification prompt word to obtain a single risk identification result.

[0140] In the embodiments of this specification, the first language model and the second language model are different. Furthermore, the first language model and the second language model have different data processing capabilities. For example, the first language model excels at data reasoning tasks, while the second language model excels at text description generation tasks. The first prompt word and the second prompt word can be different types of prompt words, the first recognition task and the second recognition task can be different types of tasks, and the first single risk identification result and the second single risk identification result can be different types of identified risk identification results.

[0141] In practical applications, the correspondence between risk identification prompt words and the large language model can be determined in advance. For different types of risk identification prompt words, different types of base large models can be used to determine the risk identification results.

[0142] In the embodiments of this specification, the second agent may generate N single risk identification results for N risk identification prompt words, and determine a risk identification result for the insurance product based on the N single risk identification results. The N single risk identification results may include a first single risk identification result and a second single risk identification result.

[0143] In the embodiments of this specification, for different types of risk identification prompt words, large language models with different functions can be called to process the recognition tasks corresponding to the risk identification prompt words, so as to assign the recognition tasks corresponding to different types of risk identification prompt words to large language models with different functions. In this way, through the diversion processing method, the recognition task corresponding to each risk identification prompt word can be processed by the large language model suitable for processing the recognition task, so as to improve the accuracy of the risk identification results.

[0144] Optionally, the risk identification result may include a factor identification result and a rule identification result. The factor identification result may be a factor identification result generated for the risk factors in the above-mentioned risk identification method, and the rule identification result may be a rule identification result generated by performing logical judgment based on the factor identification result.

[0145] In order to improve the credibility of the risk identification results output by the large speech model, the large language model can output the basis for generating the risk identification results at the same time as outputting the risk identification results.

[0146] Optionally, the risk identification result may also include an information slice determined from the insurance application information to support the risk identification result; after obtaining the risk identification result for the insurance application information determined by the large language model, it may also include: outputting the information slice and the risk identification result to the user terminal.

[0147] In the embodiments of this specification, the risk identification results may also include information slices determined from the insurance application information to support the factor identification results; these information slices are provided along with the factor identification results to a user, who may be a risk manager. For example, if the risk factor is "whether the terms and conditions mention any occupational restrictions on the policyholder," and the factor identification result is "yes," the information slice supporting the factor identification result may refer to information regarding occupational restrictions on the policyholder mentioned in the terms and conditions.

[0148] In actual applications, after generating the risk identification results, the risk identification server can feed back the risk identification results to the risk management server, which in turn feeds them back to the business server, and finally feeds them back to the user terminal corresponding to the business server, and then display the risk identification results to the user on the user terminal.

[0149] In the embodiments of this specification, when the risk identification results for insurance products are displayed to the user, the basis (information slice) used to support the risk identification results will also be displayed to the user, thereby improving the credibility of the risk identification results.

[0150] In the embodiments of this specification, a specific embodiment is proposed for a method of identifying whether there is a risk of disclosure information in insurance application information.

[0151] Optionally, the risk identification rules are used to identify the information disclosure risk in the insurance application information; the risk factors include verification trigger factors and a first verification factor that should be verified when the verification trigger factor meets predetermined conditions; the first chain of thought information is specifically used to describe the process of identifying the verification trigger factor and obtaining the trigger factor identification result; the second chain of thought information is specifically used to describe the process of judging whether the identification of the first verification factor should be performed based on the trigger factor identification result to obtain the trigger judgment result; the first chain of thought information is also used to describe the process of identifying the first verification factor and obtaining the first verification factor identification result if the trigger judgment result indicates that the identification of the first verification factor should be performed; the second chain of thought information is also used to describe the process of performing logical judgment based on the first verification factor identification result to determine the rule identification result.

[0152] In the examples of this specification, information disclosure risk may refer to the risk caused by incomplete, inaccurate, or untimely disclosure of insurance information. For example, if the terms and conditions mention occupational restrictions for the insured, but do not include a list of occupations, then this indicates that the terms and conditions involve information disclosure risk.

[0153] In order to facilitate those skilled in the art to understand this solution, the above example 1 is continued as an example for description.

[0154] In Example 1, the verification trigger factor may refer to "whether the terms and insurance instructions mention occupational restrictions on the insured". The first verification factor may refer to "whether the disclaimer mentions the occupational list / which specific occupations can be insured / which specific occupations cannot be insured, whether the terms and insurance instructions mention the occupational list / which specific occupations can be insured / which specific occupations cannot be insured, whether the health notice mentions the occupational list / which specific occupations can be insured / which specific occupations cannot be insured." The trigger factor identification result may include "yes" or "no". The trigger judgment result may include "if the trigger factor identification result is yes, then... or if the trigger factor identification result is no, then...". Each first verification factor identification result may include "yes" or "no". The rule identification result may include "no risk" or "risky".

[0155] The first chain of thought information may include the process information used in step 1 to describe the identification of the verification trigger factor and obtain the trigger factor identification result, such as "Please go to the terms and conditions or insurance policy instructions to check whether there are any occupational restrictions on the insured." The second chain of thought information may include the process information used in step 1 to describe the process information used to determine whether the first verification factor should be identified based on the trigger factor identification result and obtain the trigger judgment result, such as "If there are no restrictions, this rule will not be triggered; if the terms and conditions or insurance policy instructions mention occupational restrictions on the insured, continue the judgment."

[0156] The first chain of thought information may further include process information used in step 2 to describe if the triggering judgment result indicates that the identification of the first verification factor should be performed, then the first verification factor is identified to obtain the first verification factor identification result, such as "Please go to the disclaimer to find out whether the occupation list is mentioned / which specific occupations can be insured / which specific occupations cannot be insured, please go to the terms / insurance instructions to find out whether the occupation list is mentioned / which specific occupations can be insured / which specific occupations cannot be insured, please go to the health notice to find out whether the occupation list is mentioned / which specific occupations can be insured / which specific occupations cannot be insured." The second chain of thought information may further include process information used in step 3 to describe the logical judgment based on the first verification factor identification result to determine the rule identification result. For example, "If the disclaimer mentions the occupation list / which specific occupations can be insured / which specific occupations cannot be insured, the terms and conditions / insurance instructions mention the occupation list / which specific occupations can be insured / which specific occupations cannot be insured, and the health notice mentions the occupation list / which specific occupations can be insured / which specific occupations cannot be insured, then the rule recognition result output is no risk; if the disclaimer does not mention the occupation list / which specific occupations can be insured / which specific occupations cannot be insured, and / or the terms and conditions / insurance instructions do not mention the occupation list / which specific occupations can be insured / which specific occupations cannot be insured, and / or the health notice does not mention the occupation list / which specific occupations can be insured / which specific occupations cannot be insured, then the rule recognition result output is risky, and outputs information about what materials are missing, what statements are missing, or what is missing in the statement."

[0157] Optionally, the verification trigger factor may include trigger file type information and trigger content information, and the first verification factor may include verification file type information and verification content information. For example, if the verification trigger factor is "whether the terms and conditions and insurance information mention occupational restrictions for the policyholder," the trigger file type information may refer to "terms and conditions and insurance information," and the trigger content information may refer to "policyholder occupational restrictions." If the first verification factor is "whether the disclaimer mentions a list of occupations / which specific occupations are insurable / which specific occupations are not insurable," the verification file type information may refer to "disclaimer," and the verification content information may refer to "list of occupations / which specific occupations are insurable / which specific occupations are not insurable."

[0158] Correspondingly, the first chain of thought information can be specifically used to describe the process of identifying whether the trigger content information exists in the file represented by the trigger file type information; the first chain of thought information can also be used to describe the process of identifying whether the verification content information exists in the file represented by the verification file type information if the trigger content information exists in the file represented by the trigger file type information; the second chain of thought information can also be used to describe the process of obtaining a rule identification result indicating the presence of disclosure risk in the insurance application information if the verification content information does not exist in the file represented by the verification file type information. If the verification content information exists in the file represented by the verification file type information, the current rule verification passes; if the verification content information does not exist in the file represented by the verification file type information, the current rule verification fails.

[0159] In the embodiments of this specification, for information disclosure risks in insurance products, if the first insurance application contains the first disclosure information, the second insurance application can be identified as containing the second disclosure information. For example, for accident insurance products, if the policy details mention occupational restrictions, the policy details, such as the policy details or disclaimer, can be identified as containing a list of occupations, thereby improving the compliance of the insurance product.

[0160] In the embodiments of this specification, another specific embodiment is proposed for a method of identifying whether there is a risk of disclosure information in insurance application information.

[0161] Optionally, the risk identification rule is used to identify the information disclosure risk in the insurance application information; the risk factor includes a second verification factor; the second verification factor includes first file type information and first verification content information; the first thought chain information is specifically used to describe the process of identifying whether the first verification content information exists in the file represented by the first file type information; the second thought chain information is specifically used to describe the process of obtaining the first rule identification result indicating that there is an information disclosure risk in the insurance application information if the first verification content information does not exist in the file represented by the first file type information.

[0162] In the embodiments of this specification, the second verification factor is used to describe information that should appear in the insurance application information. For example, for medical insurance scenarios, information related to the insurance waiting period must be included in the insurance application notice. The second verification factor may include first file type information and first verification content information. The first file type information may refer to the file type of the insurance application information, such as the types corresponding to the insurance application notice, terms, health notices, and disclaimers in medical insurance. The first verification content information may refer to information that needs to be verified in the insurance application information, such as verifying whether there is relevant content about the medical waiting period in the medical insurance.

[0163] In an embodiment of this specification, the first chain of thought information can be specifically used to describe the process of identifying whether the first verification content information exists in the file represented by the first file type information. For example, for the medical insurance scenario, the first chain of thought information can be used to describe the process of identifying whether there is a medical waiting period in the insurance instructions.

[0164] In the embodiments of this specification, the second chain of thought information can be specifically used to describe the process of obtaining a first rule identification result indicating that there is an information disclosure risk in the insurance application information if the first verification content information does not exist in the file represented by the first file type information. For example, for the medical insurance scenario, the second chain of thought information can be used to describe the process of obtaining a rule identification result indicating that there is an information disclosure risk in the medical insurance (excluding the medical waiting period) if there is no medical waiting period content in the insurance application information.

[0165] In the embodiments of this specification, the information disclosure risk in insurance products can be identified by identifying whether there is a risk of missing information disclosure in the insurance application information. For example, for medical insurance products, it can be identified whether there is a medical waiting period in the insurance application instructions, thereby improving the compliance of the insurance products.

[0166] In the embodiments of this specification, a specific embodiment is also proposed for a method of identifying whether there is a risk of non-standard information in insurance application information.

[0167] Optionally, the risk identification rule is used to identify the information compliance risk in the insurance application information; the risk factor includes a third verification factor; the third verification factor includes second file type information and second verification content information; the first chain of thought information is specifically used to describe the process of identifying whether the second verification content information exists in the file represented by the second file type information; the second chain of thought information is specifically used to describe the process of obtaining the second rule identification result indicating that there is an information compliance risk in the insurance application information if the second verification content information exists in the file represented by the second file type information.

[0168] In the embodiments of this specification, information compliance risk refers to the situation where the information quality does not meet the standards or the compliance is insufficient due to failure to comply with established laws, regulations, standards or specifications during the information disclosure or processing, which may bring about legal, reputational or operational risks.

[0169] In the embodiments of this specification, the third verification factor can be used to describe information that should not appear in the insurance application information. For example, for health insurance scenarios, there should be no relevant information that restricts the insured's salary level in the insurance application notice. The third verification factor may include second file type information and second verification content information. The second file type information may refer to the file type of the insurance application information, such as: the file types corresponding to the insurance application notice, terms, health notices, disclaimers, etc. in health insurance. The second verification content information may refer to information that needs to be verified in the insurance application information, such as verifying whether there is any irregular content in the health insurance that restricts the insured's salary level.

[0170] In practical applications, for example, for health insurance products, the first chain of thought information can be used to identify whether the insurance policy contains irregularities that restrict the insured's salary level. The second chain of thought information can be used to describe the process of obtaining a second rule identification result indicating the presence of information compliance risk in the health insurance product if the insurance policy contains irregularities that restrict the insured's salary level.

[0171] In the embodiments of this specification, the information compliance risk in insurance products can be identified by identifying whether there is a risk of non-standard information in the insurance application information. For example, for health insurance products, it can be identified whether there is any non-standard content in the insurance policy that restricts the insured's salary level, thereby improving the compliance of the insurance product.

[0172] In the embodiments of this specification, a specific embodiment is also proposed for a method of identifying whether there is an information inconsistency risk in insurance application information.

[0173] Optionally, the risk identification rule is used to identify the information consistency risk in the insurance application information; the risk factor includes a fourth verification factor; the fourth verification factor includes third file type information and fourth file type information; the first thought chain information is specifically used to describe the process of identifying whether there is information consistent with the target information in the second file represented by the fourth file type information for the target information in the first file represented by the third file type information, and obtaining a consistency identification result; the second thought chain information is specifically used to describe the process of obtaining the third rule identification result indicating that there is an information consistency risk in the insurance application information if the consistency identification result indicates that there is no information consistent with the target information in the second file for the target information in the first file.

[0174] In the embodiments of this specification, information consistency risk refers to the risk caused by the inability to maintain stability and consistency of information due to inconsistencies in information content, format, standards, etc. during the disclosure of information of insurance products.

[0175] In the embodiments of this specification, the fourth check factor can be used to describe information that must be consistent across different insurance application documents. For example, in an accident insurance scenario, if the policyholder information discloses that the insured location includes location A, then the disclaimer must also disclose that the insured location includes location A. The first and second files can be different types of files. For example, if the first file is the policyholder information, the second file can be a file other than the policyholder information, such as a disclaimer.

[0176] In practical applications, for example, for accident insurance products, the first chain of thought information can be used to describe the process of identifying whether there is information consistent with the target information (the insured location includes location A) in the second file (disclaimer) for the target information (the insured location includes location A) in the first file (insurance instructions), and obtaining a consistency identification result. The second chain of thought information can be used to describe the process of obtaining a third rule identification result indicating that there is an information consistency risk in the insurance information if the consistency identification result indicates that the insured location includes location A in the insurance instructions, but there is no information consistent with the insured location including location A in the disclaimer (for example, the insured location does not include location A).

[0177] In the embodiments of this specification, information consistency risk in insurance products can be addressed by identifying whether there is a risk of information inconsistency between the first and second files in the insurance application information. For example, for accident insurance products, the risk of whether the insurance location in the insurance application instructions is consistent with the insurance location in the disclaimer can be identified, thereby improving the compliance of the insurance product.

[0178] Optionally, the first file may specifically include insurance policy information.

[0179] In the embodiments of this specification, the first document may specifically include a first disclosure document or a policy document; the second document may specifically include a second disclosure document. In practical applications, information consistency risk identification can be performed on the insurance disclosure document and the policy document, thereby preventing non-compliant disclosures in the policy document and thereby enhancing the legitimate rights and interests of the policyholder.

[0180] Optionally, the insurance disclosure information may specifically include at least one of the insurance page information, insurance instructions, insurance terms, disclaimers, health notices and special agreements.

[0181] In order to help those skilled in the art better understand the information consistency risk between insurance disclosure information and policy information, this specification provides the following specific example.

[0182] In the embodiments of this specification, for medical insurance products, if the disclaimers included in the insurance disclosure information such as the insurance application page information, insurance instructions, insurance terms, disclaimers, health notices and special agreements only mention the following contents, Disclaimer 1: The insurance company will not compensate for diseases or conditions that already exist at the time of insurance. Disclaimer 2: The insurance company is not responsible for medical events that occur during the waiting period. Disclaimer 3: The insurance company is not responsible for medical expenses caused by suicide, intentional injury, etc. However, there is no mention of Disclaimer 4: The insurance company will not pay for accidents that occur when the insured engages in high-risk activities. However, the disclaimer in the policy information generated when the customer takes out insurance includes Disclaimer 4 in addition to the above-mentioned Disclaimers 1, 2, and 3. In this scenario, there is an information consistency risk between the insurance disclosure information in the medical insurance product and the policy information. Among them, the disclaimer can be a provision that the insurance company does not bear insurance liability under certain circumstances.

[0183] Based on the embodiments of this specification, information consistency risks can be avoided by identifying inconsistent content between policy information and insurance disclosure information, thereby avoiding the risk of users being unable to claim compensation after an accident, improving product compliance, and reducing the operating risks of insurance sales platforms.

[0184] Figure 4 This is a swim lane flow diagram of a risk identification method provided by an embodiment of this specification. The business execution process of the risk identification method may involve execution entities such as a business server, a risk management server, and a risk identification server.

[0185] like Figure 4 As shown, the process may include the following steps.

[0186] Step 402: The business server sends a risk identification trigger request to the risk management server.

[0187] In the embodiment of this specification, the risk identification trigger request may be a trigger request generated for insurance information of the insurance product to be identified.

[0188] Step 404: The risk management server obtains insurance type information of the insurance product to be identified.

[0189] Step 406: Determine, based on the insurance type information, a risk identification rule applicable to risk identification of the insurance application information of the insurance product to be identified.

[0190] Step 408: The risk identification server receives the risk identification request sent by the risk management server.

[0191] In the embodiment of this specification, the risk identification request carries a rule identifier of a risk identification rule applicable to performing risk identification on the insurance application information.

[0192] Step 410: Obtain a pre-stored risk identification prompt word corresponding to the rule identifier.

[0193] In the embodiment of this specification, the risk identification prompt word is obtained by pre-writing the risk identification rule corresponding to the rule identifier.

[0194] Step 412: Input the risk identification prompt word and the insurance information into the second agent to obtain the risk identification result of the insurance product determined by the second agent by calling the large language model.

[0195] Step 414: Obtain the risk identification result for the insurance product determined by calling the large language model and returned by the second agent.

[0196] Step 416: Send the risk identification result to the risk management server.

[0197] Step 418: The risk management server sends the risk identification result fed back by the risk identification server to the business server.

[0198] Step 420: The business server sends the risk identification result fed back by the risk management server to the user terminal corresponding to the business server, so that the user terminal displays the risk identification result to the user.

[0199] Based on the same idea, the embodiments of this specification also provide a device corresponding to the above method.

[0200] Figure 5 This specification provides the corresponding embodiment Figure 2 A structural diagram of a risk identification device.

[0201] like Figure 5 As shown, the device may include:

[0202] Insurance application information acquisition module 502, used to acquire insurance application information; the insurance application information at least includes insurance disclosure information;

[0203] The prompt word acquisition module 504 is configured to acquire pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for the risk identification rule and thought chain information reflecting the risk identification logic of the risk identification rule; the thought chain information includes first thought chain information describing the factor identification result obtained by identifying the risk factor from the insurance application information, and second thought chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result;

[0204] The risk identification module 506 is configured to input the risk identification prompt words and the insurance application information into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

[0205] based on Figure 5 The embodiments of this specification also provide some improved implementation plans of the device, which are described below.

[0206] Optionally, the device may further include:

[0207] The risk identification request receiving module is used to receive a risk identification request sent by a risk management server; the risk identification request carries a rule identifier of a risk identification rule applicable to risk identification of the insurance application information.

[0208] The prompt word acquisition module 504 may specifically include:

[0209] The risk identification prompt word acquisition unit is used to acquire a pre-stored risk identification prompt word corresponding to the rule identifier; the risk identification prompt word is obtained by rewriting the risk identification rule corresponding to the rule identifier.

[0210] Optionally, the device may further include:

[0211] The insurance type information acquisition module is used for the risk management server to obtain the insurance type information of the insurance product corresponding to the insurance application information.

[0212] The risk identification rule determination module is used to determine the risk identification rules applicable to the insurance application information for risk identification based on the insurance type information.

[0213] Optionally, the device may further include:

[0214] A risk identification trigger request receiving module is used for the risk management server to receive a risk identification trigger request sent by a business server; the risk identification trigger request includes at least one of a first risk identification trigger request for triggering the identification of product risks of insurance products to be released and a second risk identification trigger request for triggering the identification of product risks of insurance products that have been released.

[0215] Optionally, the device may further include:

[0216] The risk identification rule acquisition module is used to obtain risk identification rules for identifying target risk information in the insurance application information.

[0217] The risk identification rule rewriting module is used to rewrite the risk identification rule to obtain risk identification prompt words.

[0218] Optionally, the risk identification rule acquisition module may specifically include:

[0219] A rule configuration operation receiving unit is used to receive a user's rule configuration operation for target risk information; the rule configuration operation includes a first configuration operation for configuring the risk factor to be identified, and a second configuration operation for configuring the logical judgment process executed for the risk factor.

[0220] The risk identification rule generating unit is configured to generate a risk identification rule for identifying the target risk information in response to the rule configuration operation.

[0221] Optionally, the risk identification rule acquisition module may specifically include:

[0222] The target risk information acquisition unit is used to acquire the target risk information provided by the user.

[0223] A rule generation prompt word acquisition unit is used to obtain rule generation prompt words; the rule generation prompt words include rule generation task information, rule structure information and rule case information; the rule generation task information is used to instruct the first intelligent agent to refer to the rule case information to rewrite the target risk information into a risk identification rule that conforms to the structure represented by the rule structure information.

[0224] The first input unit is used to input the target risk information and the rule generation prompt word into the first intelligent agent to obtain the risk identification rule output by the first intelligent agent.

[0225] Optionally, the risk identification prompt words specifically include N risk identification prompt words; N is a positive integer.

[0226] Optionally, the risk identification module 506 may specifically include:

[0227] The second input unit is used to assemble the N risk identification prompt words into assembled prompt words for input to the second intelligent agent; the assembled prompt words include the N risk identification prompt words and intelligent agent task information; the intelligent agent task information is used to instruct the second intelligent agent to determine N single risk identification results corresponding to the N risk identification prompt words, and determine the overall risk identification result based on the N single risk identification results; one single risk identification result corresponds to one risk identification prompt word, and different single risk identification results correspond to different risk identification prompt words.

[0228] The third input unit is used to input the assembly prompt word and the insurance information into the second agent, and obtain the risk identification result of the insurance product determined by the second agent by calling the large language model.

[0229] Optionally, the third input unit may specifically include:

[0230] An input subunit is used to input the assembly prompt word and the insurance policy information into the second intelligent agent; the second intelligent agent calls the first large language model to perform a first recognition task corresponding to the first prompt word among the N risk identification prompt words, and obtains a first single risk identification result; the second intelligent agent calls the second large language model to perform a second recognition task corresponding to the second prompt word among the N risk identification prompt words, and obtains a second single risk identification result; obtains a risk identification result for the insurance product determined by the second intelligent agent based on the N single risk identification results; the N single risk identification results include the first single risk identification result and the second single risk identification result.

[0231] Optionally, the risk identification result includes the factor identification result and the rule identification result.

[0232] Optionally, the risk identification result also includes information slices determined from the insurance application information to support the risk identification result.

[0233] The device may further include:

[0234] The risk identification result output module is used to output the information slice and the risk identification result to the user terminal.

[0235] Optionally, the risk identification rules are used to identify the information disclosure risk in the insurance application information; the risk factors include verification trigger factors and a first verification factor that should be verified when the verification trigger factor meets predetermined conditions; the first chain of thought information is specifically used to describe the process of identifying the verification trigger factor and obtaining the trigger factor identification result; the second chain of thought information is specifically used to describe the process of judging whether the identification of the first verification factor should be performed based on the trigger factor identification result to obtain the trigger judgment result; the first chain of thought information is also used to describe the process of identifying the first verification factor and obtaining the first verification factor identification result if the trigger judgment result indicates that the identification of the first verification factor should be performed; the second chain of thought information is also used to describe the process of performing logical judgment based on the first verification factor identification result to determine the rule identification result.

[0236] Optionally, the risk identification rule is used to identify the information disclosure risk in the insurance application information; the risk factor includes a second verification factor; the second verification factor includes first file type information and first verification content information; the first thought chain information is specifically used to describe the process of identifying whether the first verification content information exists in the file represented by the first file type information; the second thought chain information is specifically used to describe the process of obtaining the first rule identification result indicating that there is an information disclosure risk in the insurance application information if the first verification content information does not exist in the file represented by the first file type information.

[0237] Optionally, the risk identification rule is used to identify the information compliance risk in the insurance application information; the risk factor includes a third verification factor; the third verification factor includes second file type information and second verification content information; the first chain of thought information is specifically used to describe the process of identifying whether the second verification content information exists in the file represented by the second file type information; the second chain of thought information is specifically used to describe the process of obtaining the second rule identification result indicating that there is an information compliance risk in the insurance application information if the second verification content information exists in the file represented by the second file type information.

[0238] Optionally, the risk identification rule is used to identify the information consistency risk in the insurance application information; the risk factor includes a fourth verification factor; the fourth verification factor includes third file type information and fourth file type information; the first thought chain information is specifically used to describe the process of identifying whether there is information consistent with the target information in the second file represented by the fourth file type information for the target information in the first file represented by the third file type information, and obtaining a consistency identification result; the second thought chain information is specifically used to describe the process of obtaining the third rule identification result indicating that there is an information consistency risk in the insurance application information if the consistency identification result indicates that there is no information consistent with the target information in the second file for the target information in the first file.

[0239] Optionally, the first file specifically includes insurance policy information.

[0240] Optionally, the insurance disclosure information specifically includes at least one of the insurance page information, insurance instructions, insurance terms, disclaimers, health notices and special agreements.

[0241] Based on the same idea, the embodiments of this specification also provide devices corresponding to the above methods.

[0242] Figure 6 This specification provides the corresponding embodiment Figure 2 A schematic diagram of the structure of a risk identification device. Figure 6 As shown, the device 600 may include:

[0243] at least one processor 620; and,

[0244] A memory 610 in communication with the at least one processor; wherein,

[0245] The memory 610 stores instructions that can be executed by the at least one processor 620, and the instructions are executed by the at least one processor 620 to enable the at least one processor 620 to: obtain insurance application information; the insurance application information includes at least insurance disclosure information; obtain pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for the risk identification rules and thinking chain information reflecting the risk identification logic of the risk identification rules; the thinking chain information includes first thinking chain information describing the factor identification result obtained by identifying the risk factor from the insurance application information, and second thinking chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result; the risk identification prompt words and the insurance application information are input into the large language model to obtain the risk identification result for the insurance application information determined by the large language model.

[0246] Figure 6 The structure block diagram of a risk identification device 600 according to an embodiment of the present application is shown. The components of the device 600 include but are not limited to a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and a database 650 is used to store data.

[0247] Device 600 also includes an access device 640 that enables device 600 to communicate via one or more networks 660 .

[0248] In one embodiment of the present application, the above components of the device 600 and Figure 6 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 6 The device structure block diagram shown is for illustrative purposes only and is not intended to limit the scope of the present application. Those skilled in the art may add or replace other components as needed.

[0249] When the processor 620 executes the computer instructions, the steps of the risk identification method are implemented.

[0250] The above is a schematic diagram of a risk identification device of this embodiment. It should be noted that the technical solution of this device and the technical solution of the above-mentioned risk identification method are based on the same concept. For details not described in detail in the technical solution of the device, please refer to the description of the technical solution of the above-mentioned risk identification method.

[0251] Figure 6Examples of the network 660 include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 740 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0252] Device 600 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). Device 600 may also be a mobile or stationary server.

[0253] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device and equipment embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments. The devices, equipment and methods provided in the embodiments of this specification correspond to each other, so the devices and equipment also have beneficial technical effects similar to the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the corresponding devices and equipment will not be repeated here.

[0254] The foregoing description describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0255] In the 1990s, technological improvements could be clearly distinguished as either hardware improvements (for example, improvements to circuit structures such as diodes, transistors, and switches) or software improvements (improvements to process flows). However, with the advancement of technology, many process flow improvements today can now be considered direct improvements to hardware circuit structures. Designers almost always create the corresponding hardware circuit structure by programming the improved process flow into the hardware circuit. Therefore, it cannot be said that a process flow improvement cannot be implemented using hardware modules. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is an integrated circuit whose logical function is determined by user programming. Designers can "integrate" a digital system onto a PLD through their own programming, without having to hire a chip manufacturer to design and manufacture a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly done using "logic compiler" software. This is similar to the software compiler used when developing programs. Before compilation, the original code must also be written in a specific programming language, called a hardware description language (HDL). There is not just one HDL, but many, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art will also understand that by simply programming the method flow in one of these hardware description languages and then programming it into an integrated circuit, a hardware circuit that implements the logic method flow can be easily obtained.

[0256] The controller can be implemented in any suitable manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051 F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that in addition to implementing the controller in a purely computer-readable program code format, it is entirely possible to implement the same functionality in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules that implement the method and structures within the hardware component.

[0257] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0258] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0259] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0260] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0261] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0262] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0263] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0264] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0265] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0266] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0267] The present application may 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. The present application may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0268] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A risk identification method, applied to a risk identification server, comprising: Obtain insurance information; The insurance application information at least includes insurance disclosure information; Obtaining pre-stored risk identification prompt words for the insurance application information; The risk identification prompt includes rule description information for the risk identification rule and thought chain information reflecting the risk identification logic of the risk identification rule; the thought chain information includes first thought chain information describing the factor identification result obtained by identifying the risk factor from the insurance application information, and second thought chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result; The risk identification prompt word and the insurance application information are input into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

2. The method according to claim 1, before obtaining the pre-stored risk identification prompt words for the insurance application information, further comprising: Receiving a risk identification request sent by a risk management server; The risk identification request carries a rule identifier of a risk identification rule applicable to risk identification of the insurance application information; The obtaining of the pre-stored risk identification prompt words for the insurance application information specifically includes: Obtaining a pre-stored risk identification prompt word corresponding to the rule identifier; The risk identification prompt word is obtained by rewriting the risk identification rule corresponding to the rule identifier.

3. The method according to claim 2, before receiving the risk identification request sent by the risk management server, further comprising: The risk management server obtains insurance type information of the insurance product corresponding to the insurance application information; Based on the insurance type information, a risk identification rule applicable to risk identification of the insurance application information is determined.

4. The method according to claim 3, before the risk management server obtains the insurance type information of the insurance product corresponding to the insurance application information, further comprising: The risk management server receives a risk identification trigger request sent by the business server; the risk identification trigger request includes at least one of a first risk identification trigger request for triggering the identification of product risks of insurance products to be released and a second risk identification trigger request for triggering the identification of product risks of insurance products that have been released.

5. The method according to claim 1, before obtaining the pre-stored risk identification prompt words for the insurance application information, further comprising: Obtaining risk identification rules for identifying target risk information in the insurance application information; The risk identification rules are rewritten to obtain risk identification prompt words.

6. The method according to claim 5, wherein obtaining risk identification rules for identifying target risk information in the insurance application information specifically comprises: Receive user rule configuration operations for target risk information; The rule configuration operation includes a first configuration operation for configuring a risk factor to be identified, and a second configuration operation for configuring a logic judgment process executed for the risk factor; In response to the rule configuration operation, a risk identification rule for identifying the target risk information is generated.

7. The method according to claim 5, wherein obtaining risk identification rules for identifying target risk information in the insurance application information specifically comprises: Obtain target risk information provided by users; Get the rule generation prompt word; The rule generation prompt includes rule generation task information, rule structure information and rule case information; The rule generation task information is used to instruct the first agent to rewrite the target risk information into a risk identification rule that conforms to the structure represented by the rule structure information with reference to the rule case information; The target risk information and the rule generation prompt word are input into the first intelligent agent to obtain the risk identification rule output by the first intelligent agent.

8. The method of claim 1, wherein the risk identification prompt words specifically include N risk identification prompt words; N is a positive integer; and inputting the risk identification prompt words and the insurance application information into a large language model to obtain a risk identification result for the insurance application information determined by the large language model specifically includes: Assembling the N risk identification prompt words into assembled prompt words for input to the second agent; The assembly prompt words include the N risk identification prompt words and agent task information; The agent task information is used to instruct the second agent to determine N single risk identification results corresponding to the N risk identification prompt words, and determine an overall risk identification result based on the N single risk identification results; One of the single risk identification results corresponds to one of the risk identification prompt words, and different single risk identification results correspond to different risk identification prompt words; The assembly prompt word and the insurance information are input into the second agent to obtain a risk identification result for the insurance product determined by the second agent by calling a large language model.

9. The method of claim 8, wherein inputting the assembly prompt word and the insurance application information into the second agent to obtain a risk identification result for the insurance product determined by the second agent by calling a large language model specifically comprises: Inputting the assembly prompt and the insurance information into the second agent; The second agent calls the first large language model to perform a first recognition task corresponding to a first prompt word among the N risk identification prompt words, obtaining a first single risk identification result; The second agent calls the second language model to perform a second recognition task corresponding to a second prompt word among the N risk identification prompt words, obtaining a second single risk identification result; Obtain a risk identification result for the insurance product determined by the second agent based on the N single risk identification results; the N single risk identification results include the first single risk identification result and the second single risk identification result.

10. The method of claim 1, wherein: The risk identification result includes the factor identification result and the rule identification result.

11. The method according to claim 10, wherein: The risk identification result also includes information slices determined from the insurance application information to support the risk identification result; After obtaining the risk identification result of the insurance application information determined by the large language model, the method further includes: The information slice and the risk identification result are output to the user terminal.

12. The method of claim 1, wherein the risk identification rule is used to identify information disclosure risk in the insurance application information; the risk factor comprises a verification trigger factor and a first verification factor to be verified when the verification trigger factor meets a predetermined condition; The first thought chain information is specifically used to describe the process of identifying the verification trigger factor and obtaining the trigger factor identification result; The second thought chain information is specifically used to describe a process of determining whether to perform identification of the first verification factor based on the trigger factor identification result to obtain a trigger determination result; The first thought chain information is further used to describe a process of identifying the first check factor and obtaining a first check factor identification result if the trigger judgment result indicates that the first check factor should be identified; The second thought chain information is further used to describe a process of performing logical judgment based on the first check factor identification result to determine a rule identification result.

13. The method of claim 1, wherein: The risk identification rule is used to identify the information disclosure risk in the insurance application information; the risk factor includes a second verification factor; the second verification factor includes first file type information and first verification content information; The first thought chain information is specifically used to describe a process of identifying whether the first verification content information exists in the file represented by the first file type information; The second thought chain information is specifically used to describe the process of obtaining a first rule identification result indicating that there is an information disclosure risk in the insurance application information if the first verification content information does not exist in the file represented by the first file type information.

14. The method of claim 1, wherein: The risk identification rule is used to identify information compliance risks in the insurance application information; the risk factor includes a third verification factor; the third verification factor includes second file type information and second verification content information; The first thought chain information is specifically used to describe a process of identifying whether the second verification content information exists in the file represented by the second file type information; The second thought chain information is specifically used to describe the process of obtaining a second rule identification result indicating that there is an information standardization risk in the insurance application information if the second verification content information exists in the file represented by the second file type information.

15. The method of claim 1, wherein: The risk identification rule is used to identify the information consistency risk in the insurance application information; the risk factor includes a fourth verification factor; the fourth verification factor includes third file type information and fourth file type information; The first thought chain information is specifically used to describe a process of identifying, for target information in a first file indicated by the third file type information, whether there is information consistent with the target information in a second file indicated by the fourth file type information, and obtaining a consistency identification result; The second thought chain information is specifically used to describe the process of obtaining a third rule identification result indicating that there is an information consistency risk in the insurance information if the consistency identification result indicates that for the target information in the first file, there is no information consistent with the target information in the second file.

16. The method of claim 15, wherein: The first file specifically includes insurance policy information.

17. The method according to any one of claims 1 to 16, wherein The insurance disclosure information specifically includes at least one of the insurance page information, insurance instructions, insurance terms, disclaimers, health notices and special agreements.

18. A risk identification device comprising: Insurance information acquisition module, used to obtain insurance information; The insurance application information at least includes insurance disclosure information; A prompt word acquisition module, used to acquire pre-stored risk identification prompt words for the insurance application information; The risk identification prompt includes rule description information for the risk identification rule and thought chain information reflecting the risk identification logic of the risk identification rule; the thought chain information includes first thought chain information describing the factor identification result obtained by identifying the risk factor from the insurance application information, and second thought chain information describing the rule identification result obtained by performing logical judgment based on the factor identification result; The risk identification module is used to input the risk identification prompt word and the insurance application information into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

19. A risk identification device comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Obtaining insurance application information; the insurance application information at least includes insurance disclosure information; Obtaining pre-stored risk identification prompt words for the insurance application information; the risk identification prompt words include rule description information for a risk identification rule and thought chain information reflecting the risk identification logic of the risk identification rule; the thought chain information includes first thought chain information describing a factor identification result obtained by identifying risk factors from the insurance application information, and second thought chain information describing a rule identification result obtained by performing logical judgment based on the factor identification result; The risk identification prompt word and the insurance application information are input into a large language model to obtain a risk identification result for the insurance application information determined by the large language model.

Images