Insurance policy special risk identification method, device and system and medium

Abstract

The invention discloses an insurance policy special risk identification method, device and system and a medium. The method comprises the following steps: obtaining special content and reported contract terms of an insurance policy to be identified; establishing responsibility prompt information and object prompt information which are respectively used for comparing the special content and responsibility and object of the reported contract terms; inputting the special content, the responsibility prompt information, the object prompt information and the reported contract terms into a large language model to obtain special extension information corresponding to the responsibility prompt information and the object prompt information; and performing risk identification on the special extension information according to a preset identification strategy, and outputting risk information of the special content. Similarity comparison processing on the semantic level is carried out on the special content and the reported contract terms by utilizing a large language model, so that extension information in the special content is accurately and efficiently identified from the responsibility and object levels, and then targeted risk identification is carried out on the special extension information; and the risk identification efficiency and accuracy of the policy special contract are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and in particular to a method, device, system and medium for identifying risks in insurance policy contracts. Background Art

[0002] In the insurance industry, a special agreement (hereinafter referred to as a "special agreement") is a special clause in the insurance policy agreed upon by the insurer and the insured. It is the insurance policy liability content recognized by the Insurance Law that the insurance company and the customer have concluded in addition to the main insurance terms and supplementary insurance terms reported. It can be an extension, restriction or customized special description of the main insurance terms and supplementary insurance terms. In the case of litigation disputes, it is generally recognized that the legal effect of the special agreement is greater than the content of the policy terms. When there is a conflict between the two, the special agreement shall be given priority. Therefore, the special agreement in the insurance policy has an important impact on the execution of the subsequent policy performance and / or payment processes and risk control of the insurance company and the insurance customer.

[0003] In the product management of the insurance industry, special agreements generally need to be independent in the underwriting line, and the content must be reported before the policy is issued and approved by the underwriter, product department manager, business line manager, etc. to avoid compliance and compensation risks in special agreements. Therefore, the risk identification of special agreements in insurance policies currently requires professional underwriters to compare them one by one, and the risk identification efficiency is low and it is easy to miss. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a risk identification method, device, system and medium for insurance policy contracts that can be applied to financial technology or other related fields, and its main purpose is to improve the efficiency and accuracy of risk identification of insurance policy contracts.

[0005] The technical solution of the present invention is as follows:

[0006] A first aspect of the present invention provides a method for identifying risks in insurance policy special agreements, comprising:

[0007] Obtain the special content and reported contract terms of the insurance policy to be identified;

[0008] Constructing responsibility prompt information and subject matter prompt information, wherein the responsibility prompt information and subject matter prompt information are used to compare the responsibility and subject matter of the special contract content and the reported contract terms respectively;

[0009] Inputting the special contract content, liability prompt information, subject matter prompt information and the reported contract terms into the large language model to obtain special contract extension information corresponding to the liability prompt information and subject matter prompt information;

[0010] The risk of the special extension information is identified according to a preset identification strategy, and the risk information of the special content is output.

[0011] In one embodiment, the special contract content, liability prompt information, subject prompt information, and reported contract terms are input into a large language model to obtain special contract extension information corresponding to the liability prompt information and subject prompt information, including:

[0012] Input the special contract content, liability reminder information, subject matter reminder information and the specified content in the reported contract terms as input parameters into the large language model;

[0013] Performing semantic analysis and comparison on the special content and the designated content according to the responsibility prompt information by the large language model to obtain the special extended responsibility;

[0014] The special extended target is obtained by performing semantic analysis and comparison on the special content and the designated content according to the target prompt information through the large language model.

[0015] In one embodiment, the semantic analysis and comparison of the special content and the designated content by the large language model according to the responsibility prompt information to obtain the special extended responsibility includes:

[0016] Extracting a list of special obligations in the special obligations to form a first responsibility set;

[0017] Extracting the basic responsibility list and the responsibility exemption list in the reported contract terms to form a second responsibility set and a third responsibility set respectively;

[0018] Performing semantic analysis on the first responsibility set and the second responsibility set, and calculating a difference between the first responsibility set and the second responsibility set;

[0019] Performing semantic analysis on the first responsibility set and the third responsibility set, and calculating a first intersection of the first responsibility set and the third responsibility set;

[0020] The special extended liability in the special content is obtained according to the difference set and the first intersection set.

[0021] In one embodiment, the semantic analysis and comparison of the special content and the designated content according to the target prompt information by the large language model to obtain the special extension target includes:

[0022] Extracting a list of special items in the special content to form a first item set;

[0023] Extract the list of insurable objects with limited amount and the list of excluded objects in the reported contract terms to form the second object set and the third object set respectively;

[0024] Performing semantic analysis on the first label set and the second label set, and calculating a second intersection of the first label set and the second label set;

[0025] Performing semantic analysis on the first label set and the third label set, and calculating a third intersection of the first label set and the third label set;

[0026] The special extension target in the special content is obtained according to the second intersection and the third intersection.

[0027] In one embodiment, the specified contents in the reported contract terms include: the original text of the basic liability, the original text of the liability exemption, the original text of the limited insurable subject matter and the original text of the excluded subject matter.

[0028] In one embodiment, the step of performing risk identification on the special extended information according to a preset identification strategy and outputting risk information of the special content includes:

[0029] Confirming the extended liability type in the special contract content and the extended subject type in the special contract content according to the special contract extension information;

[0030] The risk level of the corresponding clause in the special contract is confirmed according to the extended liability type and subject type in the special contract.

[0031] In one embodiment, the step of confirming the risk level of the corresponding clause in the special contract according to the extended liability type and subject type in the special contract specifically includes:

[0032] When the extended liability types of the special content include liability exemption, unreported basic liability, or reported but removed basic liability, it is indicated that the corresponding liability clause is high risk;

[0033] When the extended liability types of the special content include basic liability types that have been reported and not removed from the shelves, the corresponding liability clause is prompted as general risk;

[0034] When the subject type of the special contract extension includes an excluded subject, or an unlimited insurable subject, it is indicated that the corresponding subject clause is high risk;

[0035] When the subject type of the special contract content extension includes a limited insurable subject matter with a limit, it is prompted that the corresponding subject clause is a general risk.

[0036] A second aspect of the present invention provides a risk identification device for a policy contract, comprising:

[0037] The acquisition module is used to obtain the special content of the insurance policy to be identified and the reported contract terms;

[0038] A construction module, used to construct responsibility prompt information and subject matter prompt information, wherein the responsibility prompt information and subject matter prompt information are used to compare the responsibility and subject matter of the special contract content and the reported contract terms respectively;

[0039] An extended comparison module, used for inputting the special contract content, liability prompt information, subject prompt information and reported contract terms into a large language model to obtain special contract extended information corresponding to the liability prompt information and subject prompt information;

[0040] The risk identification module is used to perform risk identification on the special extended information according to a preset identification strategy and output risk information of the special content.

[0041] A third aspect of the present invention provides a risk identification system for insurance policy specials, the system comprising at least one processor; and

[0042] a memory communicatively connected to the at least one processor; wherein,

[0043] The memory stores instructions that can be executed 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 execute the risk identification method of the above-mentioned insurance policy special agreement.

[0044] A fourth aspect of the present invention provides a non-volatile computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the one or more processors can execute the risk identification method for the above-mentioned policy special contract.

[0045] Beneficial effects: The present invention discloses a method, device, system and medium for identifying risks of insurance policy special contracts. Compared with the prior art, the embodiments of the present invention obtain the special contract content and the reported contract terms of the insurance policy to be identified; construct liability prompt information and subject prompt information, which are respectively used to compare the liability and subject of the special contract content and the reported contract terms; input the special contract content, liability prompt information, subject prompt information and the reported contract terms into a large language model to obtain the special contract extension information corresponding to the liability prompt information and the subject prompt information; perform risk identification on the special contract extension information according to a preset identification strategy, and output the risk information of the special contract content. By using a large language model to perform semantic-level similarity comparison processing on the special contract content and the reported contract terms, the extended information in the special contract content is accurately and efficiently identified from the liability and subject levels, and then the special contract extension information is targeted for risk identification, which effectively improves the efficiency and accuracy of risk identification of insurance policy special contracts. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0047] Figure 1 A flow chart of a risk identification method for a policy special contract provided in an embodiment of the present invention;

[0048] Figure 2 A flow chart of step S300 in the risk identification method for insurance policy special contracts provided in an embodiment of the present invention;

[0049] Figure 3 A flow chart of step S302 in the risk identification method for insurance policy special contracts provided in an embodiment of the present invention;

[0050] Figure 4 A flow chart of step S303 in the risk identification method for insurance policy special contracts provided in an embodiment of the present invention;

[0051] Figure 5 A flow chart of step S400 in the risk identification method for insurance policy special contracts provided in an embodiment of the present invention;

[0052] Figure 6 A schematic diagram of the functional modules of a risk identification device for a policy special contract provided by an embodiment of the present invention;

[0053] Figure 7 A schematic diagram of the hardware structure of a risk identification system for a policy contract provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. The embodiments of the present invention are described below in conjunction with the accompanying drawings.

[0055] In the insurance industry, a special agreement (hereinafter referred to as a "special agreement") is a special clause in the insurance policy agreed upon by the insurer and the insured. It is the insurance policy liability content recognized by the Insurance Law that the insurance company and the customer have concluded in addition to the main insurance terms and supplementary insurance terms reported. It can be an extension, restriction or customized special description of the main insurance terms and supplementary insurance terms. In the case of litigation disputes, it is generally recognized that the legal effect of the special agreement is greater than the content of the policy terms. When there is a conflict between the two, the special agreement shall be given priority. Therefore, the special agreement in the insurance policy has an important impact on the execution of the subsequent policy performance and / or payment processes and risk control of the insurance company and the insurance customer.

[0056] In the product management of the insurance industry, special agreements generally need to be independent in the underwriting line, and the content must be reported before the policy is issued and approved by the underwriter, product department manager, business line manager, etc. to avoid compliance and compensation risks in special agreements. Therefore, the risk identification of special agreements in insurance policies currently requires professional underwriters to compare them one by one, and the risk identification efficiency is low and it is easy to miss.

[0057] In order to solve the above problems, the present invention proposes a risk identification method for insurance policy special contracts. Figure 1 , Figure 1 A flowchart of an embodiment of the method for identifying risks in insurance policy special contracts provided by the present invention. The method for identifying risks in insurance policy special contracts provided in this embodiment is applied to a system consisting of a terminal device, a network and a server, wherein the network is a medium for providing a communication link between the terminal device and the server, which may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.; the operating system on the terminal device may include a handheld device operating system (iPhone operating system, iOS system), Android system or other operating system, and the terminal device is connected to the server through the network to achieve interaction, thereby performing operations such as receiving or sending data, and specifically may be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, portable computers and desktop servers, etc. Figure 1 As shown, the method specifically comprises the following steps:

[0058] S100. Obtain the special contents of the insurance policy to be identified and the reported contract terms.

[0059] In this embodiment, the reported contract terms include the main insurance terms, or both the main insurance terms and the supplementary insurance terms. These terms together constitute the basis of the insurance contract, clarifying the rights and obligations of the parties to the insurance contract, including the insurance company's coverage, liability limitations, and the insured's responsibilities. The main insurance terms are the core terms in the insurance contract, which stipulate the basic insurance liability and protection content. When the insured purchases an insurance product, the main insurance terms must be selected, and they constitute the basis of the insurance contract; the supplementary insurance terms are a supplement to the main insurance terms, which provide additional protection or coverage for certain specific risks; the supplementary insurance terms usually require the insured to choose to purchase according to their own needs. They are not mandatory, but can provide more comprehensive protection. Automated risk identification is achieved by obtaining the special content of the policy to be identified and the reported contract terms, thereby improving the efficiency and intelligence of risk identification.

[0060] S200: constructing responsibility prompt information and subject matter prompt information, wherein the responsibility prompt information and subject matter prompt information are used to compare the responsibility and subject matter of the special contract content and the reported contract terms, respectively.

[0061] In this embodiment, in order to realize automatic risk identification of special contract content and reported contract terms, the data processing and output rules of the large language model for input parameters are determined by constructing prompt information suitable for the large language model, that is, constructing the responsibility prompt and the subject prompt of the processing logic. Based on the structural framework of the insurance clause, the key risk points involved usually include insurance liability, insurance subject matter and liability exemption, among which insurance liability is to define the scope of liability, specifically to define the insurable liability, and provide a list and glossary in the text and interpretation. The extended liability beyond this list must be insured in the form of supplementary insurance; the insurance subject matter includes limited insurable subject matter and excluded subject matter, among which limited insurable subject matter is to define the compensation limit of high-risk subject matter, and to explain in the form of a list which subject matter needs to be agreed upon in the insurance value in the insurance policy before it can be used as the insurance subject matter; the excluded subject matter is to define the red line of the scope of the subject matter, and to explain in the form of a list which subject matter is not insurable, which belongs to the business red line content; the liability exemption is to define the red line of the scope of liability, which is the risk red line of the insurance policy, and to explain in the form of a list which causes the loss. If there is no explanation in the form of supplementary insurance, no compensation liability will be borne.

[0062] Based on the above-mentioned key risk points of liability and subject matter, when identifying risks in the special content, a comparison logic at the liability level, namely, liability prompt information (liability prompt), and a comparison logic at the subject matter level, namely, subject matter prompt information (subject matter prompt), are constructed to respectively compare the liability and subject matter of the special content and the reported contract terms, so that the large semantic model can automatically compare the special content and the reported contract terms based on the rules in the liability prompt information and the subject matter prompt information, identify risk points in the special content from the liability level and the subject matter level, and achieve comprehensive and efficient comparison of key risk points.

[0063] S300: Input the special contract content, liability prompt information, subject prompt information and reported contract terms into a large language model to obtain special contract extension information corresponding to the liability prompt information and subject prompt information.

[0064] In this embodiment, the special content is processed at the semantic level through the intelligent agent, and the intelligent agent is pre-defined as large language model (LLM) + memory (Memory) + planning (Planning) + tool use (Tools + Action). Large language model (LLM) is an artificial intelligence technology. It is based on the principle of deep learning and uses a large amount of text data to train a model that can understand and generate natural language; memory (Memory) is divided into short-term memory and long-term memory. Short-term memory is the context specified when the prompt is given, and long-term memory is the knowledge that the model itself does not have. In this embodiment, the special content and the reported contract terms need to be sorted out and then passed into the model as context together with the prompt of the processing logic, namely the responsibility prompt information and the subject prompt information; planning (Planning) is to let the large language model perform semantic and mathematical calculations on the passed-in parameters according to the set logic to obtain a result, and finally generate the answer through the specified format such as json.

[0065] In this embodiment, by inputting special content, liability prompt information, subject prompt information and reported contract terms, a large language model is used to perform semantic analysis and mathematical operations on the special content and the reported contract terms, and the special extended information corresponding to the liability prompt information and the subject prompt information is output, thereby obtaining the extended content of the reported contract terms corresponding to the special content. These extended contents are the risk points of the special terms of the insurance policy, thereby realizing automatic and efficient risk content extraction and providing reliable semantic recognition data for accurate risk identification.

[0066] In one embodiment, Figure 2 As shown, step S300 includes:

[0067] S301, inputting the special contract content, liability reminder information, subject matter reminder information and the specified content in the reported contract terms into the large language model as input parameters;

[0068] S302, performing semantic analysis and comparison on the special content and the designated content according to the responsibility prompt information by using the large language model to obtain the special extended responsibility;

[0069] S303: Perform semantic analysis and comparison on the special content and the designated content according to the target prompt information through the large language model to obtain a special extension target.

[0070] In this embodiment, in order to improve the data processing efficiency of the large language model, the parameters input into the model include special contract content, liability reminder information, subject matter reminder information and the specified content in the reported contract terms, wherein the specified content in the reported contract terms includes the basic liability text (such as: during the insurance period, due to natural disasters or accidents, the insurer shall be responsible for compensation in accordance with the provisions of this insurance contract. Among them, natural disasters refer to lightning strikes, heavy rains, floods, storms, tornadoes, hail, typhoons, hurricanes, sandstorms, blizzards, ice, sudden landslides, collapses, mudslides, sudden ground subsidence and other natural phenomena with strong destructive power that are beyond human control. Accidents refer to unpredictable and insured Unexpected events beyond human control that cause material losses, including fire and explosion), liability exemption text (such as: 1. Earthquakes, tsunamis and their secondary disasters; 2. Nuclear radiation, nuclear fission, nuclear fusion, nuclear pollution and other radioactive pollution; 3. Air pollution, land pollution, water pollution and other non-radioactive pollution, but non-radioactive pollution caused by insurance accidents is not limited to this; 4. Intrinsic or latent defects, natural wear and tear, natural loss, changes in the atmosphere (climate or temperature), changes in normal water levels or other gradual causes of the insured object, changes in the material itself, mildew, moisture, rat bites, insect bites, bird pecking, oxidation, rust, leakage, baking; 5. Various indirect losses; 6. Losses of the insured object itself caused by design errors, raw material defects or poor workmanship; 7. . Billboards, antennas, neon lights, solar energy devices and other external ancillary facilities of buildings, insured objects stored in the open air or in simple buildings, and simple buildings; 8. Losses caused by explosion of boilers and pressure vessels; 9. Losses of mechanical or electrical equipment caused by non-external forces; 10. Losses of mechanical or electrical equipment caused by improper operation or technical defects of the insured and its employees; 11. Shortages discovered during inventory; 12. Losses and expenses caused by interruption of public power supply, water supply, gas supply and other energy supply due to any reason), original text of the limited insurable object (such as: (i) precious property such as gold, silver, jewelry, diamonds, jade, jewelry, ancient coins, antiques, ancient books, ancient paintings, stamps, calligraphy and paintings, artworks, rare metals, etc.; (ii) dikes (i) weirs, sluice gates, railways, roads, culverts, tunnels, bridges, docks; (iii) equipment and materials in mines (pits); (iv) portable communication devices, portable computer equipment, portable photographic equipment and other portable devices and equipment; (v) projects that have not yet been put into use or accepted) and the original text of the excluded subject matter (such as: (i) land, mineral deposits, water resources and other natural resources; (ii) mines and pits; (iii) currency, tickets, securities and magnetic cards, integrated circuit (IC) cards and other cards with cash value; (iv) documents, account books, charts, technical data, computer software, computer data and other property whose value cannot be identified; (v) motor vehicles with public driving licenses; (vi) animals, plants and crops).

[0071] By extracting the original text of basic liability, liability exemption, limited insurable subject matter and excluded subject matter from the reported contract terms, and combining the liability prompt and subject matter prompt as model input parameters, the large language model can perform targeted semantic-level risk and similarity comparison to improve processing efficiency.

[0072] Specifically, the large language model is used to perform semantic analysis and comparison on the special content and the designated content according to the liability prompt information to obtain the special extended liability. For example, the main insurance liability may be expanded in the special content, which may pose potential operational or compliance risks; or if there is content that belongs to "liability exemption" in the special content, it will cross the liability scope red line in the main insurance and supplementary insurance clauses. Therefore, the large language model is used to perform semantic analysis and comparison at the liability level to obtain the extended liability content in the special content for subsequent risk identification.

[0073] At the same time, the large language model is also used to perform semantic analysis and comparison between the special contract content and the specified content according to the subject prompt information to obtain the special contract extension subject. For example, if there is content belonging to the "excluded subject" in the special contract content, it will cross the subject range red line in the main insurance and supplementary insurance clauses, which may bring business risks; or if there is content belonging to the "limited insurable subject" in the special contract content, there may be an extension of the insurance value, compensation amount or ratio, which will also bring business risks. Therefore, the large language model is used to perform semantic analysis and comparison at the subject level to obtain the extended subject content in the special contract content for subsequent risk identification. Based on the special contract extension information output by automatic semantic analysis and comparison, comprehensive and reliable risk identification can be carried out from different levels, effectively improving the comprehensiveness of the special contract risk identification of the insurance policy and avoiding omissions in risk identification.

[0074] In one embodiment, Figure 3 As shown, step S302 includes:

[0075] S3021, extracting the special responsibility list in the special content to form a first responsibility set;

[0076] S3022, extracting the basic responsibility list and the responsibility exemption list in the reported contract terms to form a second responsibility set and a third responsibility set respectively;

[0077] S3023, performing semantic analysis on the first responsibility set and the second responsibility set, and calculating a difference set between the first responsibility set and the second responsibility set;

[0078] S3024: perform semantic analysis on the first responsibility set and the third responsibility set, and calculate a first intersection of the first responsibility set and the third responsibility set;

[0079] S3025. Obtain the special extended liability in the special content according to the difference set and the first intersection set.

[0080] In this embodiment, based on the structural characteristics of the clauses and the key points of risk identification, a responsibility level comparison mechanism, namely responsibility prompt information (responsibility prompt), is constructed as follows:

[0081] (1) Extract the responsibilities mentioned in the special contract content to form a set A1;

[0082] (2) Extract the basic liabilities of the reported clauses to form set B1;

[0083] (3) Extract the liability exemptions from the reported clauses to form set C1;

[0084] (4) LLM evaluates whether the elements of sets A1 and B1 are similar from the perspective of semantic similarity (similar means that they are considered to describe the same "responsibility"). If they are similar, they are considered to be "the same element at the semantic level", that is, they belong to the intersection of the two sets, and so on. Finally, the difference set of sets A1 and B1 is calculated, that is, the elements that exist in A1 but not in B1. The number and specific elements in the difference set here are the number of basic liability clauses and specific extended content of the special contract content relative to the reported clauses;

[0085] (5) LLM evaluates whether the elements of sets A1 and C1 are similar from the perspective of semantic similarity (similar means that they are considered to describe the same "liability"). If they are similar, they are considered to be "the same element at the semantic level", that is, they belong to the intersection of the two sets, and so on. Finally, the number of elements and specific elements in the intersection of A1 and C1 are calculated, which is the number of excluded liabilities (i.e., red line liabilities) of the special extension, as well as the textual content of the excluded liabilities of the specific extension.

[0086] Therefore, based on the semantic analysis and mathematical calculation logic of the responsibility prompt, the special responsibility list in the special content is first extracted to form the first responsibility set; then the basic responsibility list and the responsibility exemption list in the reported contract terms are extracted to form the second responsibility set and the third responsibility set respectively; semantic analysis is performed on the first responsibility set and the second responsibility set to evaluate whether the elements in the first responsibility set and the second responsibility set belong to the same element at the semantic level, so as to determine the intersection elements of the two sets, and then the difference between the first responsibility set and the second responsibility set is calculated. Similarly, semantic analysis is performed on the first responsibility set and the third responsibility set, and the first intersection of the first responsibility set and the third responsibility set is calculated, where the difference between the first responsibility set and the second responsibility set represents the basic responsibility of the special content extended relative to the reported contract terms, and the first intersection represents the responsibility exemption content extended relative to the reported contract terms. The similarity of the responsibility terms of the special content and the reported conditions is processed at the semantic level through the large language model, and the semantic analysis results are mathematicized through the set to clarify the logic and working mechanism of the comparison, so as to accurately identify the special extended responsibility of the special content relative to the reported contract terms.

[0087] In one embodiment, Figure 4 As shown, step S303 includes:

[0088] Extracting a list of special items in the special content to form a first item set;

[0089] Extract the list of insurable objects with limited amount and the list of excluded objects in the reported contract terms to form the second object set and the third object set respectively;

[0090] Performing semantic analysis on the first label set and the second label set, and calculating a second intersection of the first label set and the second label set;

[0091] Performing semantic analysis on the first label set and the third label set, and calculating a third intersection of the first label set and the third label set;

[0092] The special extension target in the special content is obtained according to the second intersection and the third intersection.

[0093] In this embodiment, based on the structural characteristics of the clauses and the key points of risk identification, a mechanism for comparison at the subject level, namely subject prompt information (subject prompt), is constructed as follows:

[0094] (1) Extract the subjects mentioned in the special contract content to form set A2;

[0095] (2) Extract the “limited insurable objects” listed in the reported clauses to form set B2;

[0096] (3) Extract the “excluded subjects” listed in the reported terms to form set C2;

[0097] (4) LLM evaluates whether the elements of sets A2 and B2 are similar from the perspective of semantic similarity (similar means that they are deemed to describe the same "subject matter"). If they are similar, they are deemed to be "the same element at the semantic level". At the same time, for similar subjects, the restrictive descriptions of the amount and proportion of the subject matter in the sentence structure are detected and extracted, and so on. Finally, the intersection of sets A2 and B2 is calculated. The number and specific elements of the intersection here are the number of "limited insurable subjects" and the specific "subject matter" description content of the special content extended relative to the reported terms;

[0098] (5) LLM evaluates whether the elements of sets A2 and C2 are similar from the perspective of semantic similarity (similar means that they are considered to describe the same "subject matter"). If they are similar, they are considered to be "the same element at the semantic level", that is, they belong to the intersection of the two sets, and so on. Finally, the number of elements and specific elements in the intersection of A2 and C2 are calculated, that is, the number of excluded subjects of the special extension (that is, the subject matter that crosses the red line), as well as the description content of the "subject matter" of the specific extension.

[0099] Therefore, based on the semantic analysis and mathematical calculation logic of the subject prompt, the list of special subject matters in the special content is first extracted to form the first subject set; then the list of limited insurable subject matters and the list of excluded subject matters in the reported contract terms are extracted to form the second subject set and the third subject set respectively; semantic analysis is performed on the first subject set and the second subject set to evaluate whether the elements in the first subject set and the second subject set belong to the same element at the semantic level, so as to determine the second intersection of the two sets; similarly, semantic analysis is performed on the first subject set and the third subject set, and the third intersection of the first subject set and the third subject set is calculated, where the second intersection represents the content of the limited insurable subject matter of the special content extended relative to the reported contract terms, and the third intersection represents the content of the excluded subject matter extended relative to the reported contract terms, and the large language model is used to perform similarity processing on the subject terms of the special content and the reported conditions at the semantic level, and the semantic analysis results are mathematically processed by set method to accurately identify the special extended subject matter of the special content relative to the reported contract terms.

[0100] S400: Perform risk identification on the special extended information according to a preset identification strategy, and output risk information of the special content.

[0101] In this embodiment, the special extension information output by semantic analysis and comparison based on the big semantic model is subjected to risk identification according to a preset identification strategy. That is, the big language model only performs semantic similarity processing on the special content and the reported contract terms, and does not perform logical processing on the special extension information. The special extension information output based on the model is further subjected to logical judgment and risk level classification according to the preset device strategy, and different risk clauses in the special content are graded, thereby realizing risk identification and graded output of the policy special contracts, and efficiently obtaining accurately graded risk warning information.

[0102] In one embodiment, Figure 5 As shown, step S400 includes:

[0103] S401, confirming the extended liability type in the special contract content and the extended subject type in the special contract content according to the special contract extension information;

[0104] S402: confirm the risk level of the corresponding clause in the special contract according to the extended liability type and subject type in the special contract.

[0105] In this embodiment, the special extension information includes the special extension liability and the special extension subject matter. The type of liability of the special content extended relative to the reported contract terms is confirmed based on the special extension liability, and the type of liability specifically includes the liability exemption type, the unreported basic liability type, the reported but removed basic liability type, and the reported and not removed basic liability type; and the type of subject matter of the special content extended relative to the reported contract terms is confirmed based on the special extension subject matter, and the type of subject matter specifically includes the excluded subject matter, the unlimited limited insurable subject matter, and the limited limited insurable subject matter.

[0106] The risk level of the corresponding extended clause is further confirmed based on the extended liability type and subject matter type in the identified special contract content. Specifically:

[0107] When the extended liability types of the special contract content include liability exemption, unreported basic liability, or reported but removed basic liability, the corresponding liability clause is prompted as high risk. For high-risk extended liability clauses, the first color (such as red) can be used to highlight the prompt, and prompt information such as "please delete the clause" or "transfer the underwriting authority" can be output.

[0108] When the extended liability types of the special contract content include the basic liability types that have been reported and not removed from the shelves, the corresponding liability clause is prompted as general risk. For the extended liability clauses of general risks, the second color (such as orange) can be used to highlight the prompt, and prompt information such as please insure through additional insurance can be output.

[0109] When the subject type of the special contract extension includes an excluded subject or an unlimited insurable subject, the corresponding subject clause is prompted as high risk. For high-risk extended subject clauses, the first color (such as red) can be used to highlight the prompt, and prompt information such as please delete the clause or agree on the compensation limit can be output.

[0110] When the subject type of the special contract extension includes a limited insurable subject with a limit, the corresponding subject clause is prompted as a general risk. For the extended subject clause of general risk, the second color (such as orange) can be used to highlight the prompt, and prompt information such as "please review the limit" can be output.

[0111] This embodiment confirms the extended liability and subject matter in the special contract content based on the comparison results output by the large language model, efficiently and accurately matches the risk level of the corresponding clauses according to the preset risk identification strategy, and conducts comprehensive risk identification of the special contract extended information from the liability and subject matter levels, effectively improving the risk identification efficiency and accuracy of the insurance policy special contracts.

[0112] It should be noted that there is not necessarily a certain order between the above-mentioned steps. A person skilled in the art can understand, based on the description of the embodiments of the present invention, that in different embodiments, the above-mentioned steps may have different execution orders, that is, they may be executed in parallel, may be executed interchangeably, and so on.

[0113] Another embodiment of the present invention provides a risk identification device for a policy contract, such as Figure 6 As shown, the device 1 comprises:

[0114] The acquisition module 11 is used to acquire the special content of the insurance policy to be identified and the reported contract terms;

[0115] A construction module 12 is used to construct responsibility prompt information and subject matter prompt information, wherein the responsibility prompt information and subject matter prompt information are used to compare the responsibility and subject matter of the special contract content and the reported contract terms respectively;

[0116] The extended comparison module 13 is used to input the special contract content, liability prompt information, subject prompt information and the reported contract terms into the large language model to obtain the special contract extended information corresponding to the liability prompt information and subject prompt information;

[0117] The risk identification module 14 is used to perform risk identification on the special extended information according to a preset identification strategy and output risk information of the special content.

[0118] The module referred to in the present invention refers to a series of computer program instruction segments that can perform specific functions, which are more suitable for describing the risk identification execution process of insurance policy special agreements than programs. For the specific implementation methods of each module, please refer to the corresponding method embodiments above, which will not be repeated here.

[0119] In one embodiment, the extended comparison module 13 includes:

[0120] An input unit, used to input the special contract content, liability reminder information, subject matter reminder information and designated content in the reported contract terms as input parameters into the large language model;

[0121] A responsibility comparison unit, configured to perform semantic analysis and comparison on the special content and the designated content according to the responsibility prompt information through the large language model to obtain the special extended responsibility;

[0122] The subject comparison unit is used to perform semantic analysis and comparison on the special content and the designated content according to the subject prompt information through the large language model to obtain the special extension subject.

[0123] In one embodiment, the responsibility comparison unit includes:

[0124] A first responsibility extraction unit, used to extract a list of special responsibilities in the special content to form a first responsibility set;

[0125] A second responsibility extraction unit is used to extract the basic responsibility list and the responsibility exemption list in the reported contract terms to form a second responsibility set and a third responsibility set respectively;

[0126] A first responsibility calculation unit, configured to perform semantic analysis on the first responsibility set and the second responsibility set, and calculate a difference between the first responsibility set and the second responsibility set;

[0127] A second responsibility calculation unit is used to perform semantic analysis on the first responsibility set and the third responsibility set, and calculate a first intersection of the first responsibility set and the third responsibility set;

[0128] The extended responsibility output unit is used to obtain the special extended responsibility in the special content according to the difference set and the first intersection set.

[0129] In one embodiment, the target comparison unit includes:

[0130] A first subject extraction unit, used to extract a list of special subjects in the special content to form a first subject set;

[0131] A second subject extraction unit is used to extract a list of insurable subject matters with a limited amount and a list of excluded subject matters in the reported contract terms to form a second subject set and a third subject set respectively;

[0132] A first-label operation unit, configured to perform semantic analysis on the first-label set and the second-label set, and calculate a second intersection of the first-label set and the second-label set;

[0133] A second target operation unit, configured to perform semantic analysis on the first target set and the third target set, and calculate a third intersection of the first target set and the third target set;

[0134] The extension target output unit is used to obtain the special extension target in the special content according to the second intersection and the third intersection.

[0135] In one embodiment, the specified contents in the reported contract terms include: the original text of the basic liability, the original text of the liability exemption, the original text of the limited insurable subject matter and the original text of the excluded subject matter.

[0136] In one embodiment, the risk identification module 14 includes:

[0137] A type confirmation unit, used to confirm the extended liability type in the special contract content and the extended subject type in the special contract content according to the special contract extension information;

[0138] The risk level confirmation unit is used to confirm the risk level of the corresponding clause in the special content according to the extended liability type and subject type in the special content.

[0139] In one embodiment, the risk level confirmation unit is specifically used to:

[0140] When the extended liability types of the special content include liability exemption, unreported basic liability, or reported but removed basic liability, it is indicated that the corresponding liability clause is high risk;

[0141] When the extended liability types of the special content include basic liability types that have been reported and not removed from the shelves, the corresponding liability clause is prompted as general risk;

[0142] When the subject type of the special contract extension includes an excluded subject, or an unlimited insurable subject, it is indicated that the corresponding subject clause is high risk;

[0143] When the subject type of the special contract content extension includes a limited insurable subject matter with a limit, it is prompted that the corresponding subject clause is a general risk.

[0144] Another embodiment of the present invention provides a risk identification system for insurance policy special contracts, such as Figure 7 As shown, the system 10 includes:

[0145] One or more processors 110 and memory 120, Figure 7 A processor 110 is used as an example for description. The processor 110 and the memory 120 may be connected via a bus or other means. Figure 7 The example of connecting through bus is taken in the following.

[0146] The processor 110 is used to complete various control logics of the system 10, and it can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, a discrete gate or transistor logic, a discrete hardware component, or any combination of these components. In addition, the processor 110 can also be any traditional processor, microprocessor or state machine. The processor 110 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a DSP and / or any other such configuration.

[0147] The memory 120 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions corresponding to the risk identification method for insurance policy special contracts in the embodiment of the present invention. The processor 110 executes various functional applications and data processing of the system 10 by running the non-volatile software programs, instructions and units stored in the memory 120, that is, implements the risk identification method for insurance policy special contracts in the above method embodiment.

[0148] The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function; the data storage area may store data created according to the use of the system 10, etc. In addition, the memory 120 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 120 may optionally include a memory remotely arranged relative to the processor 110, and these remote memories may be connected to the system 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0149] One or more units are stored in the memory 120, and when executed by one or more processors 110, the risk identification method of the insurance policy special contract in any of the above method embodiments is executed, for example, Figure 1 Steps S100 to S400 are shown.

[0150] An embodiment of the present invention provides a non-volatile computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the risk identification method for the insurance policy special contract in any of the above method embodiments is executed. For example, Figure 1 Steps S100 to S400 are shown.

[0151] As an example, non-volatile storage media can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) as external cache memory. By way of illustration and not limitation, RAM can be obtained in many forms such as synchronous RAM (SRAM), dynamic RAM, (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable types of memory.

[0152] In summary, the present invention discloses a method, device, system and medium for identifying risks of insurance policy special contracts, the method includes: obtaining the special contract content and the reported contract terms of the insurance policy to be identified; constructing liability prompt information and subject prompt information, which are respectively used to compare the liability and subject of the special contract content and the reported contract terms; inputting the special contract content, liability prompt information, subject prompt information and the reported contract terms into a large language model to obtain the special contract extension information corresponding to the liability prompt information and the subject prompt information; performing risk identification on the special contract extension information according to a preset identification strategy, and outputting the risk information of the special contract content. By using a large language model to perform semantic-level similarity comparison processing on the special contract content and the reported contract terms, the extended information in the special contract content is accurately and efficiently identified from the liability and subject levels, and then the special contract extension information is targeted for risk identification, which effectively improves the efficiency and accuracy of risk identification of insurance policy special contracts.

[0153] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium, and the computer program can include the processes of the above-mentioned method embodiments when executed. The storage medium can be a memory, a disk, a floppy disk, a flash memory, an optical storage device, etc.

[0154] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for identifying risks in insurance policy contracts, characterized in that: include: Obtain the special content and reported contract terms of the insurance policy to be identified; Constructing responsibility prompt information and subject matter prompt information, wherein the responsibility prompt information and subject matter prompt information are used to compare the responsibility and subject matter of the special contract content and the reported contract terms respectively; Inputting the special contract content, liability prompt information, subject matter prompt information and the reported contract terms into the large language model to obtain special contract extension information corresponding to the liability prompt information and subject matter prompt information; The risk of the special extension information is identified according to a preset identification strategy, and the risk information of the special content is output.

2. The risk identification method of the insurance policy special contract according to claim 1 is characterized in that: The step of inputting the special contract content, liability prompt information, subject prompt information, and reported contract terms into a large language model to obtain special contract extension information corresponding to the liability prompt information and subject prompt information includes: Input the special contract content, liability reminder information, subject matter reminder information and the specified content in the reported contract terms as input parameters into the large language model; Performing semantic analysis and comparison on the special content and the designated content according to the responsibility prompt information by the large language model to obtain the special extended responsibility; The special extended target is obtained by performing semantic analysis and comparison on the special content and the designated content according to the target prompt information through the large language model.

3. The risk identification method of the insurance policy special contract according to claim 2 is characterized in that: The semantic analysis and comparison of the special content and the designated content according to the responsibility prompt information by the large language model to obtain the special extended responsibility includes: Extracting a list of special obligations in the special obligations to form a first responsibility set; Extracting the basic responsibility list and the responsibility exemption list in the reported contract terms to form a second responsibility set and a third responsibility set respectively; Performing semantic analysis on the first responsibility set and the second responsibility set, and calculating a difference between the first responsibility set and the second responsibility set; Performing semantic analysis on the first responsibility set and the third responsibility set, and calculating a first intersection of the first responsibility set and the third responsibility set; The special extended liability in the special content is obtained according to the difference set and the first intersection set.

4. The risk identification method of the insurance policy special contract according to claim 2 is characterized in that: The step of performing semantic analysis and comparison on the special content and the designated content according to the target prompt information by the large language model to obtain the special extension target includes: Extracting a list of special items in the special content to form a first item set; Extract the list of insurable objects with limited amount and the list of excluded objects in the reported contract terms to form the second object set and the third object set respectively; Performing semantic analysis on the first label set and the second label set, and calculating a second intersection of the first label set and the second label set; Performing semantic analysis on the first label set and the third label set, and calculating a third intersection of the first label set and the third label set; The special extension target in the special content is obtained according to the second intersection and the third intersection.

5. The risk identification method for insurance policy special contract according to any one of claims 1 to 4, characterized in that: The specified contents in the reported contract terms include: the original text of basic liability, the original text of liability exemption, the original text of limited insurable subject matter and the original text of excluded subject matter.

6. The risk identification method of the insurance policy special contract according to claim 1 is characterized in that: The step of performing risk identification on the special extended information according to a preset identification strategy and outputting risk information of the special content includes: Confirming the extended liability type in the special contract content and the extended subject type in the special contract content according to the special contract extension information; The risk level of the corresponding clause in the special contract is confirmed according to the extended liability type and subject type in the special contract.

7. The risk identification method of the insurance policy special contract according to claim 6 is characterized in that: The step of confirming the risk level of the corresponding clause in the special contract according to the extended liability type and subject type in the special contract specifically includes: When the extended liability types of the special content include liability exemption, unreported basic liability, or reported but removed basic liability, it is indicated that the corresponding liability clause is high risk; When the extended liability types of the special content include basic liability types that have been reported and not removed from the shelves, the corresponding liability clause is prompted as general risk; When the subject type of the special contract extension includes an excluded subject, or an unlimited insurable subject, it is indicated that the corresponding subject clause is high risk; When the subject type of the special contract content extension includes a limited insurable subject matter with a limit, it is prompted that the corresponding subject clause is a general risk.

8. A risk identification device for a policy agreement, characterized in that: include: The acquisition module is used to obtain the special content of the insurance policy to be identified and the reported contract terms; A construction module, used to construct responsibility prompt information and subject matter prompt information, wherein the responsibility prompt information and subject matter prompt information are used to compare the responsibility and subject matter of the special contract content and the reported contract terms respectively; An extended comparison module, used for inputting the special contract content, liability prompt information, subject prompt information and reported contract terms into a large language model to obtain special contract extended information corresponding to the liability prompt information and subject prompt information; The risk identification module is used to perform risk identification on the special extended information according to a preset identification strategy and output risk information of the special content.

9. A risk identification system for insurance policy specials, characterized in that: The system includes 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 execute the risk identification method of the policy special contract described in any one of claims 1-7.

10. A non-volatile computer-readable storage medium, characterized in that: The non-volatile computer-readable storage medium stores computer-executable instructions, which, when executed by one or more processors, enable the one or more processors to execute the risk identification method for the policy special clause as described in any one of claims 1-7.

Images