Insurance policy claim settlement method and device based on large language model, equipment and medium
By using large language models and insurance knowledge graphs, we can automatically process claims information and build a policy rule base, enabling efficient and accurate judgment of policy claims and solving the problem of complex and time-consuming traditional claims processes.
Patent Information
- Application Number
- CN202510838433.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-20
- Publication Date
- 2025-09-23
AI Technical Summary
The insurance claims process is complex and time-consuming, and traditional manual review is inefficient, resulting in a poor customer experience.
Using a large language model and insurance field knowledge graph, through optical character recognition and semantic parsing technology, it automatically identifies and processes claims information, builds a policy rule library, and performs logical reasoning and matching based on these technologies to generate claims judgment results.
It improves the efficiency and accuracy of claims processing, reduces errors in manual review, and enables a faster and more transparent claims process.
Smart Images
Figure CN120689148A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of artificial intelligence technology and is applicable to financial and medical scenarios. In particular, it relates to a method, device, equipment and medium for insurance claims settlement based on a large language model. Background Art
[0002] The insurance claims process has always been one of the most complex and time-consuming aspects of the insurance industry. This complexity is reflected not only in the extensive manual review process but also in the accurate interpretation and judgment of various claim documents, policy terms, and loss assessments. With the continuous development and digital transformation of the insurance claims industry, insurance companies face an increasing number of and more complex claims demands and scenarios. However, traditional manual claims processing presents numerous pain points, significantly impacting service efficiency and customer experience.
[0003] With the increasing diversification of insurance claims scenarios, the claims materials and policy terms required for different types of insurance products vary significantly during the claims process. For example, in financial scenarios, claims requirements for products such as auto insurance and property insurance vary, and claims adjusters rely on detailed analysis of the information provided by the customer to determine whether the conditions for compensation are met. Meanwhile, in medical scenarios, the claims process often requires processing a large amount of medical documents, diagnostic reports, and medical records. Especially when it comes to major illnesses or surgeries, claims adjusters must not only understand complex medical technical details but also make accurate judgments based on the insurance terms. However, in the traditional claims process, the review work is highly dependent on manual experience, and claims adjusters need to spend a lot of time and energy processing various materials for a single claim case. Problems such as long claims processing cycles and opaque compensation calculation methods often lead to customer dissatisfaction, which in turn affects the customer satisfaction of insurance companies.
[0004] In view of this, how to improve the efficiency of insurance claims processing has become a technical problem that needs to be solved urgently. Summary of the Invention
[0005] The present invention provides a large language model-based insurance claim processing method, device, equipment and medium to solve the technical problem of low efficiency in existing insurance claim processing.
[0006] In a first aspect, the present invention provides a method for settling insurance claims based on a large language model, comprising:
[0007] Obtaining the user's claim information, identifying the text information of the user's claim information using an optical character recognition algorithm, and performing semantic analysis on the identified text information of the claim information using a large language model;
[0008] Obtaining insurance policy information associated with the user's claim information, and building an insurance policy rule library based on the insurance policy information associated with the user's claim information;
[0009] Based on the large language model, the policy clauses in the policy rule base are semantically parsed to generate a set of claim rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria.
[0010] The claim rule set is semantically enhanced based on the insurance field knowledge graph, and the parsed claim information is matched with the semantically enhanced claim rule set, and the claim judgment result is generated according to the matching result.
[0011] In a second aspect, the present invention provides an insurance claim settlement device based on a large language model, comprising:
[0012] An acquisition module is used to obtain the user's claim information, identify the text information of the user's claim information through an optical character recognition algorithm, and perform semantic analysis on the identified text information of the claim information through a large language model;
[0013] A construction module, configured to obtain policy information associated with the user's claim information, and to construct a policy rule library based on the policy information associated with the user's claim information;
[0014] The parsing module is used to perform semantic analysis of each policy clause in the policy rule base based on a large language model, generating a set of claims rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria;
[0015] The matching module is used to semantically enhance the claim rule set based on the insurance field knowledge graph, match the parsed claim information with the semantically enhanced claim rule set, and generate a claim judgment result based on the matching result.
[0016] In a third aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned insurance policy claims settlement method based on a large language model are implemented.
[0017] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-mentioned insurance policy claims method based on a large language model.
[0018] The above-mentioned insurance policy claim settlement method, device, equipment and medium based on the large language model, in the implemented scheme, can obtain the user's claim information through the client, identify the text information of the user's claim information through the optical character recognition algorithm, and perform semantic analysis on the text information of the identified claim information through the large language model; obtain the insurance policy information associated with the user's claim information, and build an insurance policy rule base based on the insurance policy information associated with the user's claim information; perform semantic analysis on each insurance policy clause in the insurance policy rule base based on the large language model, and generate a claim rule set including claim scope definition, compensation ratio calculation and deductible determination criteria; perform semantic enhancement on the claim rule set based on the insurance field knowledge graph, and match the parsed claim information with the semantically enhanced claim rule set, and generate a claim determination result based on the matching result. In the present invention, semantic analysis of claim information and policy clauses through the large language model can ensure the consistency and accuracy of the understanding of claim information and policy clauses, and reduce the errors that may occur during manual review. In addition, semantic enhancement through insurance knowledge graphs can more accurately define the scope of claims, compensation ratios, and deductible determination criteria, effectively improving the processing efficiency of insurance claims. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0020] Figure 1 1 is a schematic diagram of an application environment of a method for settling insurance claims based on a large language model in one embodiment of the present invention;
[0021] Figure 2 This is a flow chart of a method for settling insurance claims based on a large language model in one embodiment of the present invention;
[0022] Figure 3 yes Figure 2 A schematic flow chart of a specific implementation of step S10;
[0023] Figure 4 yes Figure 2 A schematic flow chart of a specific implementation of step S20;
[0024] Figure 5 yes Figure 2 A schematic flow chart of a specific implementation of step S40;
[0025] Figure 61 is a schematic structural diagram of an insurance claim settlement device based on a large language model in one embodiment of the present invention;
[0026] Figure 7 is a structural diagram of a computer device in one embodiment of the present invention;
[0027] Figure 8 FIG. 2 is another structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.
[0029] The insurance claim settlement method based on the large language model provided by the embodiment of the present invention can be applied in the following areas: Figure 1 In the application environment, Figure 1 This is a schematic diagram of an application environment for a policy claim settlement method based on a large language model in one embodiment of the present invention; wherein, the client communicates with the server through a network. The server can obtain the user's claim information through the client, identify the text information of the user's claim information through an optical character recognition algorithm, and perform semantic analysis on the identified text information of the claim information through a large language model; obtain the policy information associated with the user's claim information, and build a policy rule base based on the policy information associated with the user's claim information; perform semantic analysis on each policy clause in the policy rule base based on the large language model, and generate a claim rule set including claim scope definition, compensation ratio calculation, and deductible determination criteria; perform semantic enhancement on the claim rule set based on the insurance field knowledge graph, and match the parsed claim information with the semantically enhanced claim rule set, and generate a claim determination result based on the matching result. In the present invention, semantic analysis of claim information and policy clauses through a large language model can ensure the consistency and accuracy of the understanding of claim information and policy clauses, and reduce the errors that may occur during manual review. In addition, semantic enhancement through insurance domain knowledge graphs can more accurately define the scope of claims, compensation ratios, and deductible determination criteria, effectively improving the efficiency of insurance policy claim processing. The present invention is described in detail below through specific embodiments.
[0030] See also Figure 2 As shown, Figure 2 A flowchart of a large language model-based insurance claim settlement method according to an embodiment of the present invention is provided. The large language model-based insurance claim settlement method specifically includes the following steps:
[0031] S10: Obtain the user's claim information, identify the text information of the user's claim information through an optical character recognition algorithm, and perform semantic analysis on the identified text information of the claim information through a large language model. Specifically, in the embodiment of the present invention, automatically identifying the user's claim information through an optical character recognition algorithm can significantly reduce the time and errors of manual data entry. Combining the semantic analysis of the identified text information with a large language model not only improves the speed of information processing, but also ensures accurate understanding of the information. Specifically, Figure 3 As stated, Figure 3 This is a flow chart of a specific implementation of step S10, which specifically includes the following steps S11-S13:
[0032] S11: Obtain the user's multimodal claim document, which includes at least one type of image file, scanned file and PDF file. Specifically, in an embodiment of the present invention, the claim document provided by the user usually includes images, scanned files and PDF files, and information can be extracted from different sources to ensure comprehensive coverage of all possible claim materials. For example, in a financial scenario, when a customer submits a claim document, he or she may provide photos of the accident scene, the owner's manual and insurance policy documents, and a repair list. In a medical scenario, the patient may provide a PDF diagnosis report issued by the hospital, a scanned receipt of a pathological examination, and photos of the surgical process.
[0033] S12: The text area of the multimodal claim document is located by an optical character recognition algorithm, and the located text area is subjected to character segmentation processing. Specifically, in an embodiment of the present invention, the optical character recognition algorithm can accurately identify and locate the text area in the file, whether it is a scanned document, text in an image, or embedded text in a PDF, and can accurately locate and extract it. For example, in financial auto insurance claims, photos of the accident scene may contain text information such as license plate numbers, accident descriptions, and loss conditions. The optical character recognition algorithm can extract text from images and perform character segmentation, for example, extracting key information such as license plate numbers. In medical scenarios, medical records may contain handwritten information such as doctor's diagnoses, drug prescriptions, or surgery dates. The optical character recognition algorithm can recognize these handwritten contents and perform segmentation processing to extract relevant text for subsequent claims analysis.
[0034] S13: Perform feature extraction on the text area after character segmentation processing to generate text information of a structured multimodal claim file. Specifically, in an embodiment of the present invention, the text after character segmentation may contain different character types, formats and contents. Through feature extraction technology, key data in the text can be identified and structured to facilitate subsequent analysis and decision-making. For example, in financial auto insurance claims, key information such as "accident date", "loss amount", and "repair costs" can be extracted and converted into a structured data format, such as JSON or database records, to facilitate subsequent claims approval and decision-making. In medical insurance claims, text information processed by optical character recognition can be extracted and structured to facilitate comparison of insurance terms, calculation of compensation ratios and costs.
[0035] In one embodiment of the present invention, after the text information of the user's claim information is recognized by the optical character recognition algorithm, the recognized text information of the claim information needs to be parsed by the large language model, which specifically includes the following steps S14-S15:
[0036] S14: Perform entity recognition on the text information of the structured multimodal claim document and extract the key claim elements of the multimodal claim document. Specifically, in an embodiment of the present invention, through entity recognition, the semantic content of the document can be automatically understood, and the unstructured or semi-structured text can be further refined into standardized fields to facilitate subsequent rule judgment and decision processing. For example, in financial claims scenarios, such as auto insurance or property insurance, entity recognition can quickly extract key information such as license plate number, accident time, location of the accident, name of the reporter, repair costs, etc., to provide an accurate basis for determining whether the claim conditions are met. In medical claims scenarios, entity recognition can extract data such as patient name, disease name, admission and discharge time, medical item details, invoice costs, etc.
[0037] S15: Based on the preset claim rule base, the key claim elements are checked for compliance, and the key claim elements that are not compliant are filtered. Specifically, in an embodiment of the present invention, by comparing the key claim elements with the rule base, it can be determined whether each item meets the compensation conditions, and the non-compliant data can be automatically filtered. For example, in financial claims, if it is identified that the repair costs are higher than the prescribed upper limit, it can be directly marked as an overpayment and a prompt can be given. In medical claims, if it is identified that the expense details include non-medical insurance items such as health products, daily necessities, etc., additional materials can be required according to the rules to avoid invalid compensation.
[0038] S20: Obtain the policy information associated with the user's claim information, and build a policy rule base based on the policy information associated with the user's claim information. Specifically, in the embodiment of the present invention, after semantically parsing the text information of the identified claim information through the large language model in step S10, it is also necessary to build a policy rule base based on the policy information associated with the user's claim information. Specifically, Figure 4 As stated, Figure 4 This is a flow chart of a specific implementation of step S20, which specifically includes the following steps S21-S22:
[0039] S21: Extract the original policy data from the associated policy information and parse the key rule elements of the original policy data. Specifically, in an embodiment of the present invention, the original policy content can be deeply analyzed to extract the rule elements that have an important impact on the claim, such as the insurance amount, coverage, deductible, compensation ratio, insurance period, etc. For example, in the scenario of insurance claims, by automatically extracting this information, the insurance company can more quickly evaluate whether the claim request meets the conditions and quickly make a claim decision. In the field of medical insurance, when a patient submits a claim request, it is necessary to extract specific coverage items, compensation ratios, hospitalization days limit, deductibles, etc. from the medical insurance policy. By extracting these key information from the policy, it is possible to determine whether the patient's medical expenses comply with the insurance regulations and initiate the subsequent claims process.
[0040] S22: Convert the key rule elements of the original insurance policy data into structured data, and establish a rule mapping relationship between the user's claim information and the structured key rule elements. Specifically, in an embodiment of the present invention, the information in the insurance policy can be converted into structured data so as to be compared and verified with the claim information submitted by the user. For example, in the financial auto insurance claim process, the insurance amount, deductible, compensation ratio and other information can be structured and compared with the loss caused by the accident to automatically determine the amount of compensation. In the field of medical insurance, the insurance terms that have clear compensation ratios for certain specific diseases or treatment items can be structured to accurately calculate the amount of compensation that the patient should receive after being structured, reducing the complexity of manual review.
[0041] S30: Based on the large language model, semantic analysis is performed on each policy clause in the policy rule base to generate a claim rule set that includes claim scope definition, compensation ratio calculation, and deductible determination criteria. Specifically, in the embodiment of the present invention, the terms of different insurance products vary greatly. The large language model can automatically parse different types of policy clauses and generate a special rule set based on the claim requirements to ensure that each claim case can be processed according to the correct terms. Specifically, the following steps are included in S31:
[0042] S31: Based on the Transformer model, semantic analysis is performed on the claim scope definition, compensation ratio calculation and deductible determination standards of each policy clause in the policy rule base to generate claim rules containing claim scope definition, compensation ratio calculation and deductible determination standards. Specifically, in an embodiment of the present invention, semantic analysis of the clauses in the policy rule base based on the Transformer model can convert complex insurance clauses into structured and easy-to-understand rules, thereby promoting the automation of the claims process. For example, in the financial auto insurance claims scenario, accurate claim scope definition, compensation ratio calculation and deductible determination can ensure the rapid and accurate execution of claims processes such as auto insurance and property insurance. In the field of medical insurance, the model can help parse complex medical clauses, ensure the effective implementation of rules such as medical expenses, compensation ratios and deductibles, and make the claims process more fair, transparent and efficient.
[0043] S40: Based on the insurance field knowledge graph, the claim rule set is semantically enhanced, and the parsed claim information is matched with the semantically enhanced claim rule set, and the claim judgment result is generated according to the matching result. In the embodiment of the present invention, the knowledge graph can structure the relevant knowledge in the insurance field, so that the claim rule set is not only limited to the terms themselves, but can also be semantically enhanced according to the relevant knowledge, so that the judgment is more accurate. Specifically, Figure 5 As stated, Figure 5 This is a flow chart of a specific implementation of step S40, which specifically includes the following steps S41-S44:
[0044] S41: The claim rule set is semantically enhanced through the insurance field knowledge graph to obtain a semantically enhanced claim rule set. For example, in an embodiment of the present invention, through the knowledge graph, the model can obtain a deeper level of understanding and reasoning ability, so that the original rules do not just stay at the surface text matching, but can understand the actual meaning and logic behind these rules. For example, in financial auto insurance claims, the insurance terms may stipulate "losses caused by collision" or "losses caused by natural disasters". Through the enhancement of the knowledge graph, it can be understood that "collision" may refer to vehicle collisions in accidents, and "natural disasters" can be expanded to disasters such as floods, earthquakes, and typhoons. In medical insurance claims, the rules may include expressions such as "disease treatment" or "hospitalization". Through knowledge graph enhancement, it can be identified that "disease treatment" can cover different types of diseases, such as cancer, diabetes, heart disease, etc., and further accurately process various claims requests.
[0045] S42: The semantic matching degree between the parsed claim information and the semantically enhanced claim rule set is calculated through the graph neural network model, and the calculated semantic matching degree is compared with the preset semantic matching degree threshold. Specifically, in the embodiment of the present invention, the graph neural network model can understand the relationship between the claim information and the rules and capture its semantic details. The graph neural network model can learn the semantic matching degree between the claim information and the semantically enhanced claim rule set from these text information. For example, in financial auto insurance claims, the graph neural network can identify the semantic similarity between "traffic accident" in the claim request and "full responsibility for the car accident" in the insurance terms. In medical insurance, the claim information may describe "malignant tumor surgery", while the insurance terms stipulate that "the treatment of malignant tumors is eligible for compensation." The graph neural network model can analyze the matching degree between the claim information and the rules and identify the semantic matching degree between the claim information and the rules.
[0046] S43: When the calculated semantic matching degree is greater than the preset semantic matching degree threshold, the judgment result of granting the claim is output. Specifically, in an embodiment of the present invention, when the calculated semantic matching degree is higher than the preset threshold, it means that there is a high degree of consistency between the claim information and the rules, and the claim conditions are met, and the judgment result of "granting the claim" will be output. For example, in the financial auto insurance claim scenario, if the accident involved in the claim application meets the "full-fault accident" in the terms and conditions, and the matching degree exceeds the threshold, the compensation ratio and compensation amount will be automatically determined, and the claim will be directly processed. In medical insurance claims, if a patient applies for reimbursement of medical expenses incurred due to "cancer treatment", and it fully matches the definition of "cancer treatment" in the policy terms and conditions, and the matching degree exceeds the threshold, the claim will be automatically determined to be granted.
[0047] S44: When the calculated semantic matching degree is less than or equal to the preset semantic matching degree threshold, the judgment result of claim rejection is output. Specifically, in an embodiment of the present invention, when the calculated semantic matching degree is less than or equal to the semantic matching degree threshold, it means that the degree of fit between the claim information and the rules is low, and there may be a situation where the claim conditions are not met. The judgment result of "claim rejection" is automatically output, which reduces the risk of erroneous payment and ensures the compliance of the claim. For example, in a car insurance claim, if the claim information describes "drunk driving", and the insurance rules clearly stipulate that "drunk driving will not be compensated", and the matching degree is low, the claim will be rejected. In medical claims, if a patient applies for a claim for "cosmetic plastic surgery", and the policy terms clearly state that "cosmetic plastic surgery expenses will not be paid", it will be judged that the matching degree is low and the claim will be rejected.
[0048] In one embodiment of the present invention, after generating a claim determination result, the method further includes: calculating the payout amount corresponding to the user's insurance policy based on the expense information from the parsed claim information and the payout ratio in the claim rule set; configuring interpretation rules for the claim determination, and generating a claim summary based on the configured interpretation rules. Specifically, in this embodiment of the present invention, expense information in the claim application, typically including medical expenses, repair expenses, and hospitalization expenses, can be parsed, and the extracted expenses can be calculated accordingly based on the payout ratio specified in the insurance policy. To ensure transparency and traceability of the claims process, interpretation rules for the claim determination are configured, and a claim summary is generated based on these rules, providing the user with detailed information on the claim decision. For example, in a financial scenario, suppose a user is involved in a traffic accident, with vehicle repair costs of 10,000 yuan, and the accident is entirely at fault. Based on the claim information, the vehicle repair costs and accident liability are extracted. The insurance policy stipulates that if the accident is entirely at fault, the payout ratio is 80%, resulting in a payout amount of 8,000 yuan. The claim briefing will include: the user was involved in a traffic accident and the repair cost was 10,000 yuan. According to the insurance terms, since the accident was fully responsible, the compensation ratio was 80%, so the amount to be compensated was 8,000 yuan.
[0049] As can be seen, in the above solution, semantic parsing of claim information and policy terms using a large language model ensures consistent and accurate understanding of these information and policy terms, reducing potential errors during manual review. Furthermore, semantic enhancement using the insurance domain knowledge graph allows for more precise definition of claim scope, payout ratios, and deductible criteria, effectively improving the efficiency of claim processing.
[0050] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0051] In one embodiment, a policy claim settlement device based on a large language model is provided. The policy claim settlement device based on a large language model corresponds one-to-one to the policy claim settlement method based on a large language model in the above embodiment. Figure 6 As shown, Figure 6 6 is a schematic diagram of the structure of an insurance policy claim settlement device based on a large language model according to an embodiment of the present invention. The insurance policy claim settlement device based on a large language model includes an acquisition module 61, a construction module 62, a parsing module 63, and a matching module 64. The functional modules are described in detail as follows:
[0052] An acquisition module 61 is configured to acquire a user's claim information, identify the text of the user's claim information using an optical character recognition algorithm, and perform semantic analysis on the identified text of the claim information using a large language model;
[0053] A construction module 62 is configured to obtain insurance policy information associated with the user's claim information and construct an insurance policy rule library based on the insurance policy information associated with the user's claim information;
[0054] Parsing module 63, for performing semantic parsing of each policy clause in the policy rule base based on a large language model, and generating a claim rule set including claim scope definition, compensation ratio calculation, and deductible determination criteria;
[0055] The matching module 64 is used to semantically enhance the claim rule set based on the insurance field knowledge graph, match the parsed claim information with the semantically enhanced claim rule set, and generate a claim judgment result based on the matching result.
[0056] In one embodiment, the acquisition module 61 is specifically configured to:
[0057] Obtaining a multimodal claim file of the user, wherein the multimodal claim file includes at least one type of file selected from the group consisting of an image file, a scanned file, and a PDF file;
[0058] Use optical character recognition algorithms to locate text areas in multimodal claim documents and perform character segmentation on the located text areas;
[0059] Feature extraction is performed on the text area after character segmentation to generate structured multimodal claim document text information.
[0060] In one embodiment, the acquisition module 61 is specifically configured to:
[0061] Perform entity recognition on the text information of structured multimodal claims documents and extract key claim elements from multimodal claims documents;
[0062] The key claim elements are checked for compliance based on a preset claim rule library, and key claim elements that are not compliant are filtered out.
[0063] In one embodiment, the construction module 62 is specifically configured to:
[0064] Extract original policy data from associated policy information and parse key rule elements of the original policy data;
[0065] Convert the key rule elements of the original policy data into structured data, and establish a rule mapping relationship between the user's claim information and the structured key rule elements.
[0066] In one embodiment, the parsing module 63 is specifically configured to:
[0067] Based on the Transformer model, semantic analysis is performed on the claim scope definition, compensation ratio calculation and deductible determination standards of each policy clause in the policy rule base, and claim rules containing claim scope definition, compensation ratio calculation and deductible determination standards are generated.
[0068] In one embodiment, the matching module 64 is specifically configured to:
[0069] The claim rule set is semantically enhanced through the insurance domain knowledge graph to obtain the semantically enhanced claim rule set;
[0070] The semantic matching degree between the parsed claim information and the semantically enhanced claim rule set is calculated through a graph neural network, and the calculated semantic matching degree is compared with the preset semantic matching degree threshold;
[0071] When the calculated semantic matching degree is greater than the preset semantic matching degree threshold, the judgment result of granting the claim is output;
[0072] When the calculated semantic matching degree is less than or equal to the preset semantic matching degree threshold, the judgment result of claim rejection is output.
[0073] In one embodiment, the insurance policy claims processing device based on the large language model is further used to:
[0074] Calculate the payout amount corresponding to the user's insurance policy based on the expense information of the parsed claim information and the payout ratio in the claim rule set;
[0075] Configure the interpretation rules for claim decisions and generate a claim brief based on the configured interpretation rules for claim decisions.
[0076] This invention provides a policy claims processing device based on a large language model. By using the large language model to perform semantic parsing of claim information and policy terms, it ensures consistent and accurate understanding of these information and policy terms, reducing potential errors during manual review. Furthermore, semantic enhancement through the insurance domain knowledge graph allows for more precise definition of claim scope, payout ratios, and deductible determination criteria, effectively improving the processing efficiency of policy claims.
[0077] Regarding the specific limitations of the insurance policy claims settlement device based on the large language model, please refer to the limitations of the insurance policy claims settlement method based on the large language model above, which will not be repeated here. The various modules in the above-mentioned insurance policy claims settlement device based on the large language model can be implemented in whole or in part through software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0078] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, Figure 7 1 is a schematic diagram of the structure of a computer device in one embodiment of the present invention. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the service side of a large language model-based insurance claim settlement method.
[0079] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 8 As shown, Figure 8 1 is another structural schematic diagram of a computer device in one embodiment of the present invention. The computer device includes a processor, memory, a network interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements the client-side functions or steps of a large language model-based insurance claim settlement method.
[0080] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:
[0081] Obtaining the user's claim information, identifying the text information of the user's claim information using an optical character recognition algorithm, and performing semantic analysis on the identified text information of the claim information using a large language model;
[0082] Obtaining insurance policy information associated with the user's claim information, and building an insurance policy rule library based on the insurance policy information associated with the user's claim information;
[0083] Based on the large language model, the policy clauses in the policy rule base are semantically parsed to generate a set of claim rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria.
[0084] The claim rule set is semantically enhanced based on the insurance field knowledge graph, and the parsed claim information is matched with the semantically enhanced claim rule set, and the claim judgment result is generated according to the matching result.
[0085] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:
[0086] Obtaining the user's claim information, identifying the text information of the user's claim information using an optical character recognition algorithm, and performing semantic analysis on the identified text information of the claim information using a large language model;
[0087] Obtaining insurance policy information associated with the user's claim information, and building an insurance policy rule library based on the insurance policy information associated with the user's claim information;
[0088] Based on the large language model, the policy clauses in the policy rule base are semantically parsed to generate a set of claim rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria.
[0089] The claim rule set is semantically enhanced based on the insurance field knowledge graph, and the parsed claim information is matched with the semantically enhanced claim rule set, and the claim judgment result is generated according to the matching result.
[0090] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.
[0091] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchl ink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0092] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.
[0093] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.
Claims
1. A method for insurance claim settlement based on a large language model, characterized in that: include: Obtaining the user's claim information, identifying the text information of the user's claim information using an optical character recognition algorithm, and performing semantic analysis on the identified text information of the claim information using a large language model; Obtaining insurance policy information associated with the user's claim information, and building an insurance policy rule library based on the insurance policy information associated with the user's claim information; Based on the large language model, the policy clauses in the policy rule base are semantically parsed to generate a set of claim rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria. The claim rule set is semantically enhanced based on the insurance field knowledge graph, and the parsed claim information is matched with the semantically enhanced claim rule set, and the claim judgment result is generated according to the matching result.
2. The insurance claim settlement method based on a large language model according to claim 1, characterized in that: The obtaining of the user's claim information and identifying text information of the user's claim information by an optical character recognition algorithm include: Obtaining a multimodal claim file of the user, wherein the multimodal claim file includes at least one type of file selected from the group consisting of an image file, a scanned file, and a PDF file; Use optical character recognition algorithms to locate text areas in multimodal claim documents and perform character segmentation on the located text areas; Feature extraction is performed on the text area after character segmentation to generate structured multimodal claim document text information.
3. The insurance claim settlement method based on a large language model according to claim 2 is characterized in that: The process of parsing the identified text information of the claim settlement information using the large language model includes: Perform entity recognition on the text information of structured multimodal claims documents and extract key claim elements from multimodal claims documents; The key claim elements are checked for compliance based on a preset claim rule library, and key claim elements that are not compliant are filtered out.
4. The insurance claim settlement method based on a large language model according to claim 1, characterized in that: The obtaining of the insurance policy information associated with the user's claim information and constructing an insurance policy rule library based on the insurance policy information associated with the user's claim information includes: Extract original policy data from associated policy information and parse key rule elements of the original policy data; Convert the key rule elements of the original policy data into structured data, and establish a rule mapping relationship between the user's claim information and the structured key rule elements.
5. The insurance claim settlement method based on a large language model according to claim 1, characterized in that: The large language model is used to semantically parse the various policy clauses in the policy rule base to generate a set of claim rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria, including: Based on the Transformer model, semantic analysis is performed on the claim scope definition, compensation ratio calculation and deductible determination standards of each policy clause in the policy rule base, and claim rules containing claim scope definition, compensation ratio calculation and deductible determination standards are generated.
6. The insurance claim settlement method based on a large language model according to claim 1, characterized in that: The claim rule set is semantically enhanced based on the insurance domain knowledge graph, and the parsed claim information is matched with the semantically enhanced claim rule set, and a claim determination result is generated based on the matching result, including: The claim rule set is semantically enhanced through the insurance domain knowledge graph to obtain a semantically enhanced claim rule set. The semantic matching degree between the parsed claim information and the semantically enhanced claim rule set is calculated through the graph neural network model, and the calculated semantic matching degree is compared with the preset semantic matching degree threshold; When the calculated semantic matching degree is greater than the preset semantic matching degree threshold, the judgment result of granting the claim is output; When the calculated semantic matching degree is less than or equal to the preset semantic matching degree threshold, the judgment result of claim rejection is output.
7. The insurance claim settlement method based on a large language model according to claim 1, characterized in that: After generating the claim determination result, the method further includes: Calculate the payout amount corresponding to the user's insurance policy based on the expense information of the parsed claim information and the payout ratio in the claim rule set; Configure the interpretation rules for claim decisions and generate a claim brief based on the configured interpretation rules for claim decisions.
8. An insurance claim settlement device based on a large language model, characterized in that: include: An acquisition module is used to obtain the user's claim information, identify the text information of the user's claim information through an optical character recognition algorithm, and perform semantic analysis on the identified text information of the claim information through a large language model; A construction module, configured to obtain policy information associated with the user's claim information, and to construct a policy rule library based on the policy information associated with the user's claim information; The parsing module is used to perform semantic analysis of each policy clause in the policy rule base based on a large language model, generating a set of claims rules that include claim scope definition, compensation ratio calculation, and deductible determination criteria; The matching module is used to semantically enhance the claim rule set based on the insurance field knowledge graph, match the parsed claim information with the semantically enhanced claim rule set, and generate a claim judgment result based on the matching result.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of settling insurance claims based on a large language model as described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of settling insurance claims based on a large language model as described in any one of claims 1 to 7 are implemented.