Data generation method and device based on insurance contract

Abstract

The embodiment of the specification provides a data generation method and device based on an insurance contract, wherein the method comprises the following steps: firstly, an argumentation model corresponding to the insurance contract is acquired, the insurance contract comprises a claim-related content and a non-claim-related content, the argumentation model comprises a claim path corresponding to the claim-related content and a non-claim path corresponding to the non-claim-related content, the claim path stores claim elements in the claim-related content, and the non-claim path stores non-claim elements in the non-claim-related content; then, a claim answer of a claim problem to be constructed is determined, a claim question corresponding to the claim answer is constructed based on the claim answer, the claim elements in the claim path and the non-claim elements in the non-claim path; finally, reasoning is performed on the claim question to obtain question reasoning process data, and in the case that the question construction process data and the question reasoning process data of the claim question meet consistency requirements, claim question and answer data corresponding to the insurance contract is generated according to the claim question and the claim answer.

Description

Technical Field

[0001] This document relates to the field of data processing technology, and in particular to a data generation method and apparatus based on insurance contracts. Background Technology

[0002] Data generation technology is one of the most important technologies today, with wide applications in finance, healthcare, insurance, and other fields. For example, in the insurance industry, it can be used to construct insurance claim data that simulates real users. With the diversification of insurance types, certain requirements have been placed on the construction of insurance claim data. Therefore, how to generate accurate insurance claim data is one of the problems that needs to be solved. Summary of the Invention

[0003] This specification provides one or more embodiments of a data generation method and apparatus based on insurance contracts, which can generate accurate claims Q&A data corresponding to insurance contracts.

[0004] This specification provides one or more embodiments of a data generation method based on an insurance contract, comprising: obtaining an argumentation model corresponding to the insurance contract; the insurance contract including claim-related content and claim rejection-related content; the argumentation model including a claim path corresponding to the claim-related content and a claim rejection path corresponding to the claim rejection-related content; the claim path storing claim elements in the claim-related content; the claim rejection path storing claim rejection elements in the claim rejection-related content; determining a claim answer to a claim question to be constructed; constructing a claim question corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and obtaining question construction process data for the claim question; reasoning on the claim question to obtain question reasoning process data; and, if the question construction process data and the question reasoning process data meet consistency requirements, generating claim question and answer data corresponding to the insurance contract based on the claim question and the claim answer.

[0005] This specification provides one or more embodiments of a data generation device based on an insurance contract, comprising: a model acquisition unit for acquiring an argumentation model corresponding to the insurance contract; the insurance contract includes claim-related content and claim rejection-related content; the argumentation model includes a claim path corresponding to the claim-related content and a claim rejection path corresponding to the claim rejection-related content; the claim path stores claim elements in the claim-related content; the claim rejection path stores claim rejection elements in the claim rejection-related content; a problem construction unit for determining the claim answer to be constructed, constructing a claim question corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and acquiring problem construction process data for the claim question; and a problem verification unit for reasoning on the claim question to obtain problem reasoning process data, and, if the problem construction process data and the problem reasoning process data meet consistency requirements, generating claim question and answer data corresponding to the insurance contract based on the claim question and the claim answer.

[0006] This specification provides one or more embodiments of a data generation device based on an insurance contract, comprising: a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to: acquire an argumentation model corresponding to the insurance contract; the insurance contract including claim-related content and claim rejection-related content; the argumentation model including a claim path corresponding to the claim-related content and a claim rejection path corresponding to the claim rejection-related content; the claim path storing claim elements in the claim-related content; the claim rejection path storing claim rejection elements in the claim rejection-related content; determining a claim answer to a claim question to be constructed; constructing a claim question corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and acquiring question construction process data for the claim question; reasoning about the claim question to obtain question reasoning process data; and, if the question construction process data and the question reasoning process data meet consistency requirements, generating claim question and answer data corresponding to the insurance contract based on the claim question and the claim answer.

[0007] This specification provides one or more embodiments of a computer-readable storage medium for storing computer-executable instructions, which, when executed, perform the following process: obtaining an argumentation model corresponding to an insurance contract; the insurance contract includes claim-related content and claim rejection-related content; the argumentation model includes a claim path corresponding to the claim-related content and a claim rejection path corresponding to the claim rejection-related content; the claim path stores claim elements in the claim-related content; the claim rejection path stores claim rejection elements in the claim rejection-related content; determining the claim answer to be constructed for a claim question; constructing a claim question corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and obtaining question construction process data for the claim question; reasoning about the claim question to obtain question reasoning process data; and, if the question construction process data and the question reasoning process data meet consistency requirements, generating claim question and answer data corresponding to the insurance contract based on the claim question and the claim answer.

[0008] This specification provides one or more embodiments of a computer program product, including a computer program that, when executed by a processor, performs the following process: obtaining an argumentation model corresponding to an insurance contract; the insurance contract includes claim-related content and claim rejection-related content; the argumentation model includes a claim path corresponding to the claim-related content and a claim rejection path corresponding to the claim rejection-related content; the claim path stores claim elements in the claim-related content; the claim rejection path stores claim rejection elements in the claim rejection-related content; determining the claim answer to be constructed for a claim question; constructing a claim question corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and obtaining question construction process data for the claim question; reasoning about the claim question to obtain question reasoning process data; and, if the question construction process data and the question reasoning process data meet consistency requirements, generating claim question and answer data corresponding to the insurance contract based on the claim question and the claim answer. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 A schematic diagram illustrating an application scenario of a data generation method based on an insurance contract provided in one or more embodiments of this specification; Figure 2 A flowchart illustrating a data generation method based on an insurance contract provided for one or more embodiments of this specification; Figure 3 A schematic diagram of an insurance contract argumentation model provided for one or more embodiments of this specification; Figure 4 A flowchart illustrating another data generation method based on an insurance contract provided for one or more embodiments of this specification; Figure 5 A schematic diagram illustrating the generation of claims issues based on an argumentation model, provided for one or more embodiments of this specification; Figure 6 A flowchart illustrating yet another data generation method based on an insurance contract provided for one or more embodiments of this specification; Figure 7 A schematic diagram illustrating the verification of claims issues based on an argumentation model, provided for one or more embodiments of this specification; Figure 8 A schematic diagram illustrating the data generation process of an insurance contract provided for one or more embodiments of this specification; Figure 9 This is a schematic diagram of the structure of a data generation device based on an insurance contract according to one or more embodiments of this specification; Figure 10 This is a schematic diagram of the structure of a data generation device based on an insurance contract, provided for one or more embodiments of this specification. Detailed Implementation

[0010] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0011] This specification provides one or more embodiments of a data generation method based on an insurance contract. Figure 1 This is a schematic diagram illustrating an application scenario of a data generation method based on an insurance contract provided in one or more embodiments of this specification, such as... Figure 1As shown, the scenario includes a server 101. This server 101 can be a single server, a server cluster, or a cloud server. This server 101 can execute the methods described in one or more embodiments of this specification to generate claim Q&A data corresponding to the insurance contract.

[0012] like Figure 1 As shown, in the process of generating claim Q&A data corresponding to the insurance contract, the server uses the insurance contract as input. First, it obtains the argumentation model corresponding to the insurance contract. The argumentation model is created based on the contract content and stores the payout and denial elements of the insurance contract. Then, for the claim answer to the claim question to be constructed, the server constructs the corresponding claim question based on the argumentation model. Finally, it performs reasoning on the claim question to obtain the reasoning process data. If the question construction process data and the question reasoning process data meet the consistency requirements, the server generates the claim Q&A data corresponding to the insurance contract based on the claim question and claim answer. The server can then output the claim question and claim answer that meet the consistency requirements.

[0013] Therefore, through one or more embodiments of this specification, it is possible to create a claims problem where the data from the problem construction process and the data from the problem reasoning process meet the consistency requirement. The consistency between the data from the problem construction process and the data from the problem reasoning process indicates that the construction principle of the claims problem can be reproduced during the reasoning process. Furthermore, since the claims problem is constructed based on the argumentation model corresponding to the insurance contract, and the construction principle of the claims problem reflects the claims logic of the insurance contract, it can be determined that the claims problem can reflect the claims logic of the insurance contract during the reasoning process. Therefore, based on the claims problem and the corresponding claims answer, claims question and answer data corresponding to the insurance contract can be accurately generated.

[0014] For example, in a model training scenario, the claim question and its answer can be used as training data for the model. Since the claim question can reflect the claim logic of the insurance contract during the reasoning process, training the model using the claim question and its answer can enable the model to accurately learn the claim logic of the insurance contract, allowing the trained model to accurately answer users' questions related to insurance contract claims.

[0015] Figure 2 This specification provides a flowchart illustrating a data generation method based on an insurance contract, which can be applied to and executed by a server. The server may include... Figure 1 Server 101 in the middle. For example... Figure 2 As shown, the method includes: Step S202: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content; Step S204: Determine the claim answer to the claim question to be constructed; based on the claim answer, the claim elements in the claim payment path, and the claim rejection elements in the claim rejection path, construct the claim question corresponding to the claim answer, and obtain the claim question construction process data. Step S206: Reason about the claims issue to obtain the problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, generate the claims question and answer data corresponding to the insurance contract based on the claims issue and the claims answer.

[0016] In this embodiment, the insurance contract can be any type of insurance contract, such as accident insurance, critical illness insurance, etc., and there is no limitation here. For each insurance contract, it can be... Figure 2 The methodology retrieves the claims Q&A data corresponding to the insurance contract. When there are multiple insurance contracts, each contract can be processed separately. Figure 2 The process involves obtaining the corresponding claims Q&A data for the insurance contract. The following explanation uses any insurance contract as an example.

[0017] For ease of explanation later, here is an example of a specific form of insurance contract, which can be called Insurance Contract A. The contents of Insurance Contract A are as follows: Indemnification Clause: If the insured is hospitalized due to (1) accidental injury or (2) illness after a 30-day waiting period, a hospitalization allowance shall be paid.

[0018] Disclaimer: No compensation will be paid for hospitalization caused by the following circumstances: (1) driving under the influence of alcohol; (2) undergoing cosmetic surgery (except for necessary facial repair caused by accidental injury).

[0019] Definition clause: Accidental injury refers to an external, non-disease-related sudden event.

[0020] Before step S202, an argumentation model corresponding to the insurance contract needs to be constructed. This argumentation model includes, but is not limited to, nested Toulmin models, repudiable logic argument frameworks, and Wigmore argument diagrams, which have four types of argumentation nodes (D, W, C, R) and can represent the facts required for the argument, the basis required for the argument, the conclusion being argued, and the rebuttal of the argument. Specifically, D (Data) nodes represent the facts required for the argument, W (Warrant) nodes represent the basis required for the argument, C (Claim) nodes represent the conclusion being argued, and R (Rebuttal) nodes represent the rebuttal of the argument. The relationship between the four types of argumentation nodes (D, W, C, R) can be expressed as follows: based on the facts required for the argument and the basis stated in the argument, the conclusion being argued can be derived; when there is a rebuttal, the conclusion being argued is invalid. D nodes can be called fact nodes, W nodes can be called basis nodes, C nodes can be called conclusion nodes, and R nodes can be called rebuttal nodes.

[0021] The following explanation uses a nested Thurmin model as an example of the argumentation model.

[0022] In one embodiment, the argumentation model corresponding to the insurance contract can be constructed in the following way: Based on the claim-related content in the insurance contract, construct the claim-related path corresponding to the claim-related content; Based on the content related to claim denial in the insurance contract, construct the claim denial path corresponding to the content related to claim denial. Based on the compensation path and the denial path, construct the corresponding argumentation model for the insurance contract.

[0023] First, extracting each claim-related element from the insurance contract yields one or more claim-related elements. Each claim-related element includes a claim establishment description text, and may or may not include related definition clauses. Taking insurance contract A as an example, two claim-related elements can be extracted from insurance contract A. Claim-related element 1 includes: The insured is hospitalized due to accidental injury, where accidental injury refers to an external, non-disease-related sudden event. Claim-related element 2 includes: A hospitalization allowance is paid if hospitalized due to illness after a 30-day waiting period. Specifically, claim-related element 1 includes the claim establishment description text "The insured is hospitalized due to accidental injury" and the definition clause "Accidental injury refers to an external, non-disease-related sudden event." Claim-related element 2 includes the claim establishment description text "A hospitalization allowance is paid if hospitalized due to illness after a 30-day waiting period." For each claim-related element, a corresponding claim path is constructed, thus obtaining one or more claim paths for the insurance contract.

[0024] Then, by extracting each claim-rejection-related content from the insurance contract, one or more claim-rejection-related contents can be obtained. Each claim-rejection-related content includes a description of the grounds for claim rejection, and may or may not include related definition clauses, and may or may not include related claim exemption descriptions. Taking the aforementioned insurance contract A as an example, two claim-rejection-related contents can be extracted from insurance contract A. Claim-rejection-related content 1 includes: No compensation for drunk driving. Claim-rejection-related content 2 includes: No compensation for cosmetic surgery, except for necessary facial repairs caused by accidental injury. Among them, claim-rejection-related content 1 includes the description of the grounds for claim rejection "No compensation for drunk driving". Claim-rejection-related content 2 includes the description of the grounds for claim rejection "No compensation for cosmetic surgery" and the description of the claim exemption "Except for necessary facial repairs caused by accidental injury". For each claim-rejection-related content, a claim rejection path corresponding to that content is constructed, thereby obtaining one or more claim rejection paths in the insurance contract.

[0025] In one example, the description text for the establishment of compensation can be extracted from the compensation clause, and the description text for the establishment of denial of compensation and the description text for denial of compensation exemption can be extracted from the exemption clause.

[0026] Finally, based on the payout and denial paths, an argumentation model corresponding to the insurance contract is constructed. For example, the various payout paths are combined into a payout unit, and the various denial paths are combined into a denial unit. Based on the payout and denial units, the argumentation model corresponding to the insurance contract is obtained.

[0027] In some embodiments, as described above, the insurance contract includes one or more claims-related elements; before obtaining the argumentation model corresponding to the insurance contract, the claims path corresponding to the claims-related elements is constructed through the following methods: For each claim-related item, create a claim verification node corresponding to that item, and based on the claim verification node, create a claim path corresponding to that item. The compensation justification nodes include: compensation basis node, compensation condition node, and compensation conclusion node; the compensation basis node stores the compensation basis content in the compensation-related content; the compensation condition node stores the compensation elements in the compensation-related content; and the compensation conclusion node is used to indicate the compensation.

[0028] For each extracted claim-related content, a claim path is constructed. The claim path includes claim justification nodes corresponding to the claim-related content. These nodes include: a claim basis node, a claim condition node, and a claim conclusion node. The claim basis node is the W node mentioned above, the claim condition node is the D node mentioned above, and the claim conclusion node is the C node mentioned above. The claim basis node, claim condition node, and claim conclusion node are concatenated in sequence to obtain the claim path. The concatenation order of these nodes within the claim path can be set as needed.

[0029] The Compensation Basis node stores the compensation basis content from the compensation-related content, the Compensation Condition node stores the compensation elements from the compensation-related content, and the Compensation Conclusion node is used to represent the compensation. In one example, the compensation basis content is the compensation establishment description text from the compensation-related content. Compensation elements are the elements in the compensation-related content that make the compensation valid. These elements can be located in the compensation establishment description text. The compensation establishment description text can be decomposed into atomic propositions related to compensation validity; each atomic proposition obtained from the decomposition is a compensation element. The content of the Compensation Conclusion node can include instructions or information used to represent the compensation.

[0030] For example, regarding the aforementioned claim-related content 1: the insured is hospitalized due to an accidental injury, where an accidental injury refers to an external, non-disease-related sudden event. The descriptive text "the insured is hospitalized due to an accidental injury" can be extracted from claim-related content 1 as the claim basis node. By decomposing the descriptive text "the insured is hospitalized due to an accidental injury" in claim-related content 1 into atomic propositions related to claim establishment, we obtain two atomic propositions: "accident" and "hospitalization." Therefore, we obtain two claim elements: "accident" and "hospitalization." The text "claim" is then used as the content of the claim conclusion node.

[0031] For example, regarding the aforementioned claim-related content 2: If hospitalization due to illness occurs after the 30-day waiting period, a hospitalization allowance will be paid. This claim-related text only includes a description of the claim's establishment. The description of the claim's establishment, "If hospitalization due to illness occurs after the 30-day waiting period, a hospitalization allowance will be paid," can be used as the content of the claim basis node. By decomposing the claim establishment description of claim-related content 2 into atomic propositions related to claim establishment, three atomic propositions are obtained: "after the 30-day waiting period," "illness," and "hospitalization." Therefore, three claim elements are obtained: "after the 30-day waiting period," "illness," and "hospitalization." The text "claim" is then used as the content of the claim conclusion node.

[0032] Considering that the compensation-related content may or may not include definition clauses, when the compensation-related content includes definition clauses, a B (Backing) node can be created to represent supporting evidence for the compensation. This node can be filled with the definition clauses from the compensation-related content. For compensation-related content 1, a B node can be created, with the content of the B node being "Accidental injury refers to an external, non-disease-related sudden event." It is understandable that for compensation-related content 2, there is no B node. The B node can be called a supporting evidence node.

[0033] As can be seen, this embodiment can extract the basis and elements of compensation from the compensation-related content, thereby accurately creating a compensation path.

[0034] In this embodiment, the number of compensation elements in the compensation condition node is one; or, the number of compensation elements in the compensation condition node is multiple, and there is a logical connection relationship between each compensation element.

[0035] Taking any claim-related content as an example, this claim-related content corresponds to a claim path. The claim condition node of this claim path is used to store the claim elements in the claim-related content that make the claim conclusion valid. The number of these claim elements can be one or more. When there are multiple claim elements, they are connected by logical connection relationships. The logical connection relationships between the claim elements can be determined from the claim establishment description text. The logical connection relationships include one or more of the "AND" relationship and "OR" relationship.

[0036] Taking the aforementioned claim-related content 1 as an example, its claim elements include "accident" and "hospitalization." These two elements are connected by an "AND" logical relationship. Based on this, the content in the claim condition node can be represented as " (Accident) (Hospitalization).

[0037] Taking the aforementioned claim-related content 2 as an example, its claim elements include "after the 30-day waiting period," "illness," and "hospitalization." These three elements are connected by an AND statement. Based on this, the content in the claim condition node can be represented as " (After a 30-day waiting period) (disease) (Hospitalization).

[0038] Of course, in some cases, the payout condition node in the payout path may only store one payout element. In other cases, the payout condition node in the payout path may also store multiple payout elements, connected by an "OR" relationship, or connected by both "AND" and "OR" relationships. For example, payout element 1... (Compensation Element 2) "Compensation Element 3"

[0039] It is understandable that the principle of a claims path is to arrive at a claim settlement conclusion based on the claims elements and the logical connections (if any) between them. Therefore, this embodiment not only extracts the claims elements required for a claim to be settled within a claims path, but also marks the logical connections between these elements, establishing a data foundation for generating subsequent claims questions.

[0040] In some embodiments, as described above, the insurance contract includes one or more claim denial-related clauses; before obtaining the argumentation model corresponding to the insurance contract, the claim denial path corresponding to the claim denial-related clauses is constructed through the following methods: For each claim denial-related item, create a claim denial argumentation node corresponding to the claim denial-related item, and create a claim denial path corresponding to the claim denial-related item based on the claim denial argumentation node; The denial of claim argument nodes include: denial basis node, denial condition node, and denial conclusion node; the denial basis node stores the denial basis content in the denial-related content; the denial condition node stores the denial elements in the denial-related content; and the denial conclusion node is used to indicate denial of claim.

[0041] For each extracted claim denial-related content, a claim denial path is constructed. The claim denial path includes claim denial argumentation nodes corresponding to the claim-related content. Claim denial argumentation nodes include: a claim denial basis node, a claim denial condition node, and a claim denial conclusion node. The claim denial basis node is the aforementioned W node, the claim denial condition node includes at least the aforementioned D node, and the claim denial conclusion node is the aforementioned C node. The claim denial basis node, claim denial condition node, and claim conclusion node are concatenated in sequence to obtain the claim denial path. The concatenation order of the claim denial basis node, claim denial condition node, and claim conclusion node in the claim denial path can be set as needed.

[0042] The "Rejection Basis" node stores the content supporting the rejection, while the "Rejection Condition" node stores the elements related to the rejection. The "Rejection Conclusion" node indicates the rejection. In one example, the "Rejection Basis" content includes a rejection justification description and a rejection waiver description (if applicable). Rejection elements are elements related to the rejection within the rejection content. These elements can be located in either the rejection justification description or the rejection waiver description. The rejection justification description can be decomposed into atomic propositions related to the rejection's validity; each resulting atomic proposition is a rejection element. Similarly, the rejection waiver description can be decomposed into atomic propositions related to the rejection waiver; each resulting atomic proposition is a rejection element. The "Rejection Conclusion" node may include instructions or information indicating the rejection.

[0043] Considering that the content related to the claim denial may or may not include the definition clause, when the content related to the claim denial includes the definition clause, a B (Backing) node can be created to represent the evidence for the claim denial. This node can be filled with the definition clause from the content related to the claim denial.

[0044] As can be seen, this embodiment can extract the basis and elements for claim denial from the relevant content, and then accurately create a claim denial path.

[0045] In one embodiment, the rejection condition node includes a rejection establishment node, which stores rejection establishment elements from the rejection-related content; the rejection elements include rejection establishment elements; in the rejection establishment node, the number of rejection establishment elements is one; or, in the rejection establishment node, the number of rejection establishment elements is multiple, and the various rejection establishment elements have logical connection relationships.

[0046] In this embodiment, for claim-rejection related content that does not include a description of the claim rejection waiver, the corresponding claim rejection condition node includes a claim rejection establishment node, which is also the aforementioned D node. The D node stores the claim rejection establishment elements from the claim rejection related content. In one case, the number of claim rejection establishment elements in the claim rejection establishment node is one; in another case, the number of claim rejection establishment elements in the claim rejection establishment node is multiple, and these elements are logically connected. The logical connection relationship includes one or more of the "AND" and "OR" relationships. The claim rejection establishment elements can be located in the claim rejection establishment description text. The claim rejection establishment description text can be decomposed into atomic propositions related to claim rejection establishment; each atomic proposition obtained from the decomposition is a claim rejection establishment element. The logical connection relationships between the various claim rejection establishment elements can be determined from the claim rejection establishment description text.

[0047] For example, regarding the aforementioned content related to claim denial 1: No compensation will be paid for drunk driving. This content only includes the descriptive text for establishing the claim denial. The descriptive text "No compensation will be paid for drunk driving" can be used as the content of the claim denial basis node. The decomposition of the descriptive text "No compensation will be paid for drunk driving" into atomic propositions related to the establishment of claim denial yields an atomic proposition "Drunk driving." "Drunk driving" is used as an element for establishing claim denial, and this element is used as the content of the claim denial node. The text "Rejection" is then used as the content of the claim denial conclusion node.

[0048] For an illustration of the logical connections between multiple elements that justify a claim denial, please refer to the examples of compensation elements mentioned above; they will not be repeated here.

[0049] It is understandable that, in the above situations, the principle of the claim denial path is based on the elements that establish a claim denial and the logical connections between these elements (if any), leading to the conclusion of non-payment. Therefore, through this embodiment, when the elements for claim denial include the elements that establish a claim denial, not only can the elements that establish a claim denial in a claim denial path be extracted, but the logical connections between these elements can also be marked, establishing a data foundation for generating subsequent claims issues.

[0050] In one embodiment, the claim rejection condition node includes a claim rejection establishment node and a claim rejection waiver node. The claim rejection establishment node stores the claim rejection establishment elements in the claim rejection-related content; the claim rejection waiver node stores the claim rejection waiver elements in the claim rejection-related content; the claim rejection elements include the claim rejection establishment elements and the claim rejection waiver elements. In the denial of claim establishment node, there are one or more elements for denial of claim establishment. When there are multiple elements for denial of claim establishment, there is a logical connection between each element. In the denial of claim waiver node, there are one or more elements for denial of claim waiver. When there are multiple elements for denial of claim waiver, there is a logical connection between each element.

[0051] It should be noted that the claim settlement condition nodes include the claim rejection establishment node, or both the claim rejection establishment node and the claim rejection waiver node, and will not only include the claim rejection waiver node.

[0052] In this embodiment, for claim-rejection related content containing description text of claim rejection waiver, the corresponding claim rejection condition nodes include a claim rejection establishment node and a claim rejection waiver node. The claim rejection establishment node is also the aforementioned D node, and the claim rejection waiver node is also the aforementioned R node. The D node stores the claim rejection establishment elements in the claim rejection related content. The R node stores the claim rejection waiver elements in the claim rejection related content. In one case, the number of claim rejection establishment elements in the claim rejection establishment node is one; in another case, the number of claim rejection establishment elements in the claim rejection establishment node is multiple, and these elements are logically connected. In one case, the number of claim rejection waiver elements in the claim rejection waiver node is one; in another case, the number of claim rejection waiver elements in the claim rejection waiver node is multiple, and these elements are logically connected.

[0053] Logical connections include one or more of the "AND" and "OR" relationships. Elements constituting a claim rejection can be located within the description text of the claim rejection. This description text can be decomposed into atomic propositions related to claim rejection; each resulting atomic proposition is a claim rejection element. Elements constituting a claim exemption can be located within the description text of the claim exemption. This description text can be decomposed into atomic propositions related to claim exemption; each resulting atomic proposition is a claim exemption element. The logical connections between the various elements constituting a claim rejection can be determined from the description text of the claim rejection. The logical connections between the various elements constituting a claim exemption can be determined from the description text of the claim exemption.

[0054] The above describes four scenarios: 1. There is one element for establishing a claim rejection and one element for waiver of claim rejection; 2. There are multiple elements for establishing a claim rejection, and these elements are logically connected, while there is one element for waiver of claim rejection; 3. There are multiple elements for waiver of claim rejection, and these elements are logically connected, while there is one element for establishing a claim rejection; 4. There are multiple elements for establishing a claim rejection, and these elements are logically connected, as are multiple elements for waiver of claim rejection, and these elements are logically connected.

[0055] For example, regarding the aforementioned claim denial content 2: Cosmetic surgery is not covered, except for necessary facial repairs resulting from accidental injury. This claim denial content includes the description text "Cosmetic surgery is not covered" and the description text "Necessary facial repairs resulting from accidental injury are excluded." These two texts can be used as the basis for claim denial. The description text "Cosmetic surgery is not covered" is decomposed into atomic propositions related to claim denial, resulting in the atomic proposition "Cosmetic surgery," which is used as the element for claim denial. Similarly, the description text "Necessary facial repairs resulting from accidental injury are excluded" is decomposed into the atomic proposition "Accidental facial repairs," which is used as the element for claim denial. The elements for claim denial are used as the content of the claim denial node, and the elements for claim denial exemption are used as the content of the claim exemption node. The text "Rejection of Claim" will be used as the content of the rejection of claim conclusion node.

[0056] For an illustration of the logical connection between multiple elements for establishing a claim denial, and an illustration of the logical connection between multiple elements for waiving a claim denial, please refer to the examples of compensation elements mentioned above, which will not be repeated here.

[0057] It is understandable that, in the above scenarios, the principle of the claim denial path is that, based on the elements establishing the claim denial and the logical connections between these elements (if any), a conclusion of non-payment can be reached; and based on the elements of claim exemption and the logical connections between these elements (if any), a conclusion of payment can be reached, and the conclusion of payment can refute the conclusion of non-payment. Therefore, through this embodiment, when the elements of claim denial include elements establishing the claim denial and elements of claim exemption, not only can the elements establishing the claim denial in a claim denial path be extracted, but the logical connections between each element establishing the claim denial can also be marked. Similarly, when the elements of claim exemption in a claim denial path are extracted, the logical connections between each element of claim exemption can also be marked, thus establishing a data foundation for subsequently generating claims issues.

[0058] Taking insurance contract A as an example again, Figure 3 A schematic diagram of the argumentation model of the insurance contract provided in one or more embodiments of this specification, such as Figure 3 As shown, through the above process, a claim basis node is created based on the content of the claim basis; a claim condition node is created based on the claim elements; a claim conclusion node representing the claim is created; a claim rejection basis node is created based on the content of the claim rejection basis; a claim rejection establishment node is created based on the claim rejection establishment elements; a claim rejection waiver node is created based on the claim rejection waiver elements (if any); and a claim rejection conclusion node representing the claim rejection is created. Thus, the following path is obtained: The claim path 1 corresponding to claim-related content 1 includes: claim basis node W "the insured was hospitalized due to accidental injury", claim conditions node D " (Accident) (Hospitalization), payout node C "payment" and supporting node B "accidental injury refers to an external, non-disease-related sudden event"; The claim path 2 corresponds to claim-related content 2. Claim path 2 includes: claim basis node W "If hospitalization is caused by illness after a 30-day waiting period, a hospitalization allowance will be paid", and claim condition node D " (After a 30-day waiting period) (disease) (Hospitalization), payout node C node "Payout"; The claim rejection path 1 corresponds to the relevant content of the claim rejection. Claim rejection path 1 includes: the claim rejection basis node W node "No compensation for drunk driving", the claim rejection establishment node D node " (Drunk driving), claim denial node C "claim denial"; The relevant content regarding the claim denial corresponds to claim denial path 2, which includes: the claim basis node W, "No compensation will be paid for cosmetic surgery, except for necessary facial repairs caused by accidental injury," and the claim establishment node D, "..." (Cosmetic Surgery), Node C "Rejection" and Node R "Rejection Waiver" (Unexpected facial repair).

[0059] Figure 3 In this model, each payout path forms a payout unit, and each rejection path forms a rejection unit. In each payout path, the content of the R node can be empty. In one example, using a Neural Turing Machine (NTM) model, the payout clauses, definition clauses, and exclusion clauses in insurance contract A can be extracted. The payout clauses yield a payout establishment description text, and the exclusion clauses yield a rejection establishment description text and a rejection exemption description text. Through this process, each payout path and each rejection path can be obtained. Each payout path and rejection path can also include a Q (Qualifier) ​​node. In the payout path, the Q node stores the limiting words for establishing the payout path, such as "under normal circumstances." In the rejection path, the Q node stores the limiting words for establishing the rejection path, such as "usually under normal circumstances." In the above process, P and Q represent atomic propositions. Atomic propositions can be obtained by decomposing the payout establishment description text, rejection establishment description text, and rejection exemption description text using a Large Language Model (LLM).

[0060] It is worth noting that the argumentation model created through the above process contains two nested layers. The first layer: When a denial-of-claim exemption element exists, this element can refute the denial-of-claim conclusion node, leading to a conclusion indicating compensation. For example, when the facts corresponding to both the denial-of-claim establishment element and the denial-of-claim exemption element are true, the conclusion of the denial-of-claim path is compensation. The second layer, case 1: The conclusion of each denial-of-claim path can refute the conclusion of each compensation path. For example, if one denial-of-claim path concludes with no compensation and another compensation path concludes with compensation, then the overall conclusion of the argumentation model is no compensation. The second layer, case 2: The conclusion of each compensation path can refute the conclusion of each denial-of-claim path. For example, if one compensation path concludes with no compensation and another denial-of-claim path concludes with compensation, then the overall conclusion of the argumentation model is no compensation.

[0061] After establishing the argumentation model corresponding to the insurance contract through the above process, in step S202, the argumentation model corresponding to the insurance contract is obtained, for example, by retrieving the argumentation model. In step S204, the claim answer to the claim problem to be constructed is determined. Based on the claim answer, the payout elements in the payout path, and the rejection elements in the rejection path, the claim problem corresponding to the claim answer is constructed, and the problem construction process data of the claim problem is obtained. In some embodiments, determining the claim answer to the claim problem to be constructed may be by obtaining a pre-configured claim answer. One or more claim answers may be pre-configured, and each claim answer is used to construct a claim problem.

[0062] In some embodiments, the number of payout paths is at least one, and each payout path constitutes a payout unit; the number of rejection paths is at least one, and each rejection path constitutes a rejection unit; based on the claim answer, the payout elements in the payout path, and the rejection elements in the rejection path, a claim question corresponding to the claim answer is constructed, including: Based on the claims answers, determine the first sub-claims answer for the claims settlement unit and the second sub-claims answer for the claims denial unit; Based on the first sub-claim answer, determine the first claim answer for each claim path regarding the claim issue; based on the second sub-claim answer and the configuration information regarding whether the claim rejection waiver is effective in the claim answer, determine the first rejection answer for each claim rejection path regarding the claim issue. The claims problem is constructed based on the first payout answer for each payout path, the first rejection answer for each rejection path, the payout elements in the payout path, and the rejection elements in the rejection path.

[0063] In this embodiment, after obtaining the pre-configured claim answers, firstly, based on the claim answers, a first sub-claim answer for the claim issue by the compensation unit and a second sub-claim answer for the claim issue by the denial unit are determined. The claim answer includes either compensation or no compensation; the first sub-claim answer of the compensation unit includes either compensation or no compensation, and the second sub-claim answer of the denial unit also includes either compensation or no compensation.

[0064] Then, for each claim unit, based on the first sub-claim answer, a first claim answer is determined for each claim path regarding the claim issue. The first sub-claim answer is related to the first claim answer for each claim path regarding the claim issue, and the first claim answer includes either claim or no claim. The claim answer can be configured to specify whether a claim rejection waiver is in effect. A claim rejection waiver being in effect means that for a claim rejection path, the facts corresponding to both the elements establishing the claim rejection and the elements corresponding to the claim rejection waiver are true, causing the claim rejection path to output a claim conclusion due to the claim rejection waiver being in effect. The claim answer can be configured to specify whether the claim rejection waiver is in effect or not. When in effect, it can also be configured to specify which particular claim rejection path's claim rejection waiver is in effect. For each claim rejection unit, based on the second sub-claim answer and the configuration information regarding whether the claim rejection waiver is in effect in the claim answer, a first claim rejection answer is determined for each claim rejection path regarding the claim issue. The second sub-claim answer is related to the first claim rejection answer for each claim rejection path regarding the claim issue, and the first claim rejection answer includes either claim or no claim.

[0065] Finally, based on the first payout answer for each payout path, the first rejection answer for each rejection path, the payout elements in the payout path, and the rejection elements in the rejection path, the claims problem is constructed.

[0066] Therefore, through this embodiment, based on the claim answer, the first sub-claim answer of the claim unit, the second sub-claim answer of the claim rejection unit, the first claim answer of the claim path, and the first claim rejection answer of the claim rejection path can be determined in sequence, thereby accurately constructing the claim question that can obtain the claim answer.

[0067] In some embodiments, if the claim answer specifies a first sub-claim answer for the payout unit and a second sub-claim answer for the denial unit, then the first sub-claim answer for the payout unit and the second sub-claim answer for the denial unit can be directly extracted from the claim answer.

[0068] In some embodiments, the claims answer includes the argumentation model's answer to the claims question; the above-mentioned determination of the first sub-claims answer of the payout unit and the second sub-claims answer of the denial unit based on the claims answer includes: If the answer to the claim indicates that a claim has been made, then the first sub-claim answer is determined to indicate that a claim has been made, and the second sub-claim answer is determined to indicate that a claim has been made. If the claim answer indicates no compensation, then the first sub-claim answer is determined to be compensable, and the second sub-claim answer is determined to be no compensation; or, the first sub-claim answer is determined to be no compensation, and the second sub-claim answer is determined to be no compensation; or, the first sub-claim answer is determined to be no compensation, and the second sub-claim answer is determined to be compensable.

[0069] Based on the second layer of nesting described above, the conclusion of each denial path can refute the conclusion of each payment path, and the conclusion of each payment path can refute the conclusion of each denial path. Based on this, the argumentation model can obtain a claim answer representing a claim in the following case: 1. The payment unit derives a first sub-claim answer representing a claim in the claim problem, and the denial unit derives a second sub-claim answer representing a claim in the claim problem. The argumentation model can arrive at a non-payment answer for the claim problem in any of the following three scenarios: 1. The payout unit arrives at a first sub-claim answer indicating payment, and the rejection unit arrives at a second sub-claim answer indicating non-payment; 2. The payout unit arrives at a first sub-claim answer indicating non-payment, and the rejection unit arrives at a second sub-claim answer indicating non-payment; 3. The payout unit arrives at a first sub-claim answer indicating non-payment, and the rejection unit arrives at a second sub-claim answer indicating payment. These four scenarios can be represented in Table 1 below.

[0070] Table 1

[0071] In other words, according to the voidable claim logic of insurance contracts, the argumentation model determines that a claim can be made only when both the payout unit and the denial unit are determined to be eligible for a claim; otherwise, the argumentation model determines that a claim cannot be made.

[0072] Therefore, in this step, based on the above four situations and the revocable claim logic of the insurance contract, through the above process, and based on the answer of the argumentation model to the claim problem, the first sub-claim answer of the compensation unit and the second sub-claim answer of the denial unit to the claim problem can be accurately determined.

[0073] Next, for the claims unit, after obtaining the first sub-claims answer to the claims issue, the first claims answer for each claims path can be determined based on the first sub-claims answer. In specific implementation, if the first sub-claims answer indicates a claim, at least one claims path is randomly selected and its first claims answer is set as a claim, while the first claims answer for other claims paths is set as no claim; if the first sub-claims answer indicates no claim, then the first claims answer for each claims path is determined to be no claim.

[0074] For each claim denial unit, the first denial answer for the claim issue needs to be determined based on the configuration information regarding whether the denial waiver is effective in the second sub-claim answer and the claim answer. In practice, the first denial answer for each claim issue is determined based on the configuration information regarding whether the denial waiver is effective in the second sub-claim answer and the claim answer, including: If the configuration information indicates that the claim rejection waiver is effective in the claim settlement answer, then the first claim rejection answer that determines the claim rejection path related to the claim rejection waiver indicates a claim settlement, and, based on the second sub-claim settlement answer, the first claim rejection answer that determines the claim rejection path unrelated to the claim rejection waiver is determined; If the configuration information indicates that the claim rejection waiver is not effective in the claim answer, then the first rejection answer for each rejection path is determined based on the second sub-claim answer.

[0075] As mentioned earlier, the claim answer can be configured to allow or disallow the claim waiver. When allowed, it can be further configured to specify which claim waiver applies. If the claim answer configuration indicates that the claim waiver is active, then at least one claim path with the claim waiver element is selected from among the claim paths as the claim path related to the claim waiver. This claim path can be specified in the configuration information or randomly selected. Since the claim waiver is active for this claim path, the first claim answer for this claim path is determined to be accepted. Then, based on the second sub-claim, the first claim answer for the claim path unrelated to the claim waiver is determined. The claim path unrelated to the claim waiver is any claim path other than the selected claim paths with the claim waiver element.

[0076] In one scenario, although the waiver of the aforementioned waiver of claim is in effect, if the second sub-claim answer of the waiver unit indicates no claim, then at least one waiver of claim must be selected from among the waiver of claim and the first waiver answer of that waiver of claim must be determined as no claim. In addition, the first waiver answer of the other waiver of claim must be determined as claim, thus creating a situation where although the waiver of claim is in effect, some waiver of claim does not agree to claim, and therefore the waiver unit does not agree to claim.

[0077] In another scenario, if the waiver of the aforementioned waiver for the waiver is in effect, and the second sub-claim answer of the waiver unit indicates a claim, then for waiver paths other than those for which the waiver is in effect (waiver paths unrelated to the waiver), the waiver answer for these waiver paths can be set to a claim, thus creating a situation where the waiver is in effect, all waiver paths agree to a claim, and the waiver unit agrees to a claim.

[0078] If the configuration information for the claim answer indicates that the waiver of claim denial is not effective, then the first denial answer for each denial path is determined based on the second sub-claim answer. In practice, if the second sub-claim answer indicates a claim, then the first denial answer for each denial path is determined to be a claim; if the second sub-claim answer indicates no claim, then at least one denial path is randomly selected and its first denial answer is set to no claim, and the first denial answer for the other denial paths is set to a claim.

[0079] Therefore, through this embodiment, for the claim rejection unit and claim rejection path, the configuration information of whether the claim rejection waiver is effective in the claim answer can be taken into account, and the first claim rejection answer for each claim rejection path can be accurately determined. This makes the generated claim questions cover the situation where the claim rejection waiver is effective, and makes the generated claim questions closer to the real insurance business scenario, thereby improving the refinement and accuracy of the constructed claim questions.

[0080] Taking the aforementioned insurance contract A as an example, it has four payment paths: 1, 2, 1, and 2.

[0081] Claim path 1 includes: claim basis node W "Insured hospitalized due to accidental injury", claim condition node D " (Accident) (Hospitalization), payout node C "payment" and supporting node B "accidental injury refers to an external, non-disease-related sudden event"; Claim path 2 includes: claim basis node W "hospitalization allowance will be paid if hospitalization is caused by illness after the 30-day waiting period", and claim condition node D " (After a 30-day waiting period) (disease) (Hospitalization), payout node C node "Payout"; Claim denial path 1 includes: claim denial basis node W "No compensation for drunk driving", and compensation condition node D " (Drunk driving), claim denial node C "claim denial"; Pathway 2 for claim denial includes: node W, the basis for claim denial: "No compensation will be paid for cosmetic surgery, except for necessary facial repairs caused by accidental injury"; and node D, the establishment of claim denial. (Cosmetic Surgery), Node C "Rejection" and Node "Rejection Waiver" (Unexpected facial repair).

[0082] In one example, a pre-configured claim answer is obtained, which includes: the first sub-claim answer of the claim unit for the claim question is a claim, the second sub-claim answer of the claim rejection unit for the claim question is a claim, and the claim rejection unit has a claim rejection waiver in effect without specifying the effective claim rejection path.

[0083] Through the above process, firstly, based on the first sub-claim answer, the first claim answer for claim paths 1 and 2 is determined. Since the first sub-claim answer is "claim," claim path 1 is randomly selected and its first claim answer is set to "claim," and the first claim answer for claim path 2 is set to "no claim."

[0084] Then, based on the configuration information regarding the second sub-claim answer and whether the claim rejection waiver is in effect, the first claim rejection answer for claim rejection paths 1 and 2 is determined. Since the second sub-claim answer is a claim and the claim rejection waiver is in effect, and no effective claim rejection path is specified, claim rejection path 1 is selected. Because the claim rejection waiver is in effect, the first claim rejection answer for claim rejection path 1 is determined to be a claim. Also, since the second sub-claim answer is a claim, the first claim rejection answer for claim rejection path 2 is also determined to be a claim.

[0085] It is understandable that in this example, the argumentation model's answer to the claims problem is "claims." This information can be represented in Table 2 below.

[0086] Table 2

[0087] After obtaining the first payout answer for each payout path and the first rejection answer for each rejection path, in some embodiments, a claim problem is constructed based on the first payout answer for each payout path, the first rejection answer for each rejection path, the payout elements in the payout path, and the rejection elements in the rejection path, including: For each payout path, the first state of the payout element in the payout path is determined based on the first payout answer of the payout path; the first state indicates whether the fact corresponding to the payout element is true. For each claim denial path, based on the first denial answer of the claim denial path, determine the second state of the denial element in the claim denial path; the second state indicates whether the fact corresponding to the denial element is true. The claims problem is constructed based on the first state of the payout elements in each payout path and the second state of the denial elements in each denial path.

[0088] For each payout path, based on the first payout answer for that path, the first state of the payout element in that path is determined. The first state indicates whether the fact corresponding to the payout element is true. The first state can be represented by a state value such as 0 or 1, where 1 indicates that the fact corresponding to the payout element is true, and 0 indicates that the fact corresponding to the payout element is not true.

[0089] Furthermore, for each claim rejection path, based on the first rejection answer for that path, a second state of the rejection element in that path is determined. The second state indicates whether the fact corresponding to the rejection element is true. The second state can be represented by a state value such as 0 or 1, where 1 indicates that the fact corresponding to the rejection element is true, and 0 indicates that the fact corresponding to the rejection element is not true.

[0090] Finally, based on the first state of the payout elements in each payout path and the second state of the rejection elements in each rejection path, a claims problem is constructed. The claims problem can be a text describing the user's claims story, indicating whether the facts corresponding to the payout elements in each payout path are true, and whether the facts corresponding to the rejection elements in each rejection path are true. In one example, a claims problem is constructed using a large language model based on the first state of the payout elements in each payout path and the second state of the rejection elements in each rejection path, combined with storyline expansion capabilities.

[0091] Therefore, through this embodiment, it is possible to accurately determine whether the facts corresponding to the compensation elements in each compensation path are true, and whether the facts corresponding to the denial elements in each denial path are true, based on the claims answer, thereby generating claims questions that can obtain claims answers, and improving the logical rationality and accuracy of the generated claims questions.

[0092] The following example, using any payout path, illustrates the process of determining the first state. In this step, when determining the first state of a payout element, it's necessary to consider the number of payout elements in the payout path and the logical connections between them. It's understandable that the principle of a payout path is that a claim settlement conclusion can be reached based on the payout elements and the logical connections (if any) between them.

[0093] In one scenario, the compensation path includes a compensation element. If the first compensation answer of the compensation path is to pay, then the fact corresponding to the compensation element is determined to be true, and a first state such as state value 1 is generated to indicate that the fact is true. If the first compensation answer of the compensation path is not to pay, then the fact corresponding to the compensation element is determined to be false, and a first state such as state value 0 is generated to indicate that the fact is false.

[0094] In another scenario, there are multiple payout elements in the payout path, and these elements are logically connected. Accordingly, based on the first payout answer of the payout path, the first state of each payout element in the payout path is determined, including: Based on the first payout answer in the payout path and the logical connection between each payout element in the payout path, determine the first state of each payout element in the payout path.

[0095] Specifically, the logical connections between the various compensation elements in the compensation path are determined. These logical connections include one or more of the following: "AND" and "OR" relationships. Example 1: The compensation elements in the compensation path include elements A, B, and C. At the compensation condition node, the relationship between elements A, B, and C is that all three corresponding facts are true; that is, they are logically ANDed. Example 2: The compensation elements in the compensation path include elements A, B, and C. At the compensation condition node, the relationship between elements A, B, and C is that any one of them corresponds to a true fact; that is, they are logically ORed. Example 3: The compensation elements in the compensation path include elements A, B, and C. At the compensation condition node, the relationship between elements A and B is that any one of them corresponds to a true fact. The relationship between this true fact and element C is that all corresponding facts are true, which can be represented as "(Element A... Element B) In the case of "Element C", the logical connection between the three elements includes both logical AND and logical OR relationships.

[0096] After determining the logical connections between the various compensation elements in the compensation path, the first state of each compensation element in the compensation path can be determined based on the first compensation answer and the logical connections between the various compensation elements in the compensation path. Each compensation element has a first state, which can be represented by a state value: 0 indicates that the corresponding fact is not true, and 1 indicates that the corresponding fact is true.

[0097] In Example 1 above, if the first compensation answer is compensation, then the first state of elements A, B, and C all indicate that the corresponding facts are true; for example, the state values ​​of elements A, B, and C are all 1. If the first compensation answer is no compensation, then the first state of at least one of elements A, B, and C indicates that the corresponding facts are not true; for example, the state value of element A is 0, and the state values ​​of elements B and C are 1.

[0098] In Example 2 above, if the first compensation answer is compensation, then the first state of at least one of elements A, B, and C indicates that the corresponding fact is true. For example, the state value of element A is 1, and the state values ​​of elements B and C are 0. If the first compensation answer is no compensation, then the first state of elements A, B, and C all indicate that the corresponding fact is not true. For example, the state values ​​of elements A, B, and C are all 0.

[0099] Example 3 above can be represented as "(Element A)" Element B) If the first payout answer is "pay out," then it can be determined that the first state of at least one of elements A and B indicates that the corresponding fact is true, and the first state of element C also indicates that the corresponding fact is true. For example, the state value of element A is 1, the state value of element B is 0, and the state value of element C is 1. If the first payout answer is "no payout," then it can be determined that the first states of elements A and B both indicate that the corresponding fact is false, or the first state of element C indicates that the corresponding fact is false. For example, the state values ​​of elements A and B are 0, and the state value of element C is 0.

[0100] As can be seen, through this embodiment, when there are multiple claim elements in the claim path and there are logical connections between the claim elements, the first state of each claim element in the claim path can be accurately determined based on the first claim answer of the claim path and the logical connection between the claim elements in the claim path, so that the generated claim problem is closer to the real insurance business scenario.

[0101] The following example, using any denial path, illustrates the process of determining the second state. In this step, when determining the second state of the denial factors, it is necessary to consider whether the denial path includes denial exemption factors, the number of denial-establishing factors, and the logical connections between these factors. In the above cases, the principle of the denial path is that the conclusion of non-payment can be reached based on the denial-establishing factors and their logical connections (if any), while the conclusion of compensation can be reached based on the denial exemption factors and their logical connections (if any), and the conclusion of compensation can refute the conclusion of non-payment.

[0102] Scenario 1: The claim denial process includes one element that would justify a claim denial. In this scenario, if the first rejection answer in the rejection path is "no claim," then the fact corresponding to the elements establishing the rejection is confirmed, and a second state, such as state value 1, is generated to indicate that the fact is confirmed. If the first rejection answer in the rejection path is "claim," then the fact corresponding to the elements establishing the claim is not confirmed, and a second state, such as state value 0, is generated to indicate that the fact is not confirmed.

[0103] Scenario 2: The claim denial process includes multiple elements for establishing a claim denial. In this case, the second state of each element in the claim rejection path can be determined based on the first rejection answer and the logical connection between the elements that establish the rejection in the path. This process can be referenced from the process of determining the first state of the elements when there are multiple elements, and will not be repeated here.

[0104] Scenario 3: The claim denial path includes one element for establishing a claim denial and one element for waiving a claim denial. In this case, a state value of 1 in the second state indicates that the fact is established, and a state value of 0 in the second state indicates that the fact is not established. If the first rejection answer in the rejection path is no compensation, the result can be determined as follows: the state value of the rejection establishment element is 1 and the state value of the rejection waiver element is 0. If the first rejection answer in the rejection path is compensation, the result can be determined as follows: the state value of the rejection establishment element is 1 and the state value of the rejection waiver element is 1, or the state value of the rejection establishment element is 0 and the state value of the rejection waiver element is 0, or the state value of the rejection establishment element is 0 and the state value of the rejection waiver element is 1. The above situations can be represented by Table 3 below.

[0105] Table 3

[0106] Scenario 4: The claim denial path includes one element for establishing a claim denial and multiple elements for waiver of claim denial. In this case, it is necessary to determine whether the claim waiver is effective in the claim rejection path. Based on whether the claim waiver is effective and the logical connection between the various claim waiver elements, the first sub-state of each claim waiver element is determined. Then, based on the first claim rejection answer of the claim rejection path and the first sub-state of each claim waiver element, the second sub-state of the claim rejection establishment element is determined. The second state includes both the first sub-state of the claim waiver element and the second sub-state of the claim rejection establishment element.

[0107] Scenario 5: The claim denial path includes multiple elements for establishing a claim denial and one element for waiving a claim denial. In this case, it is necessary to determine whether the claim waiver is effective in the claim rejection path. Based on whether the claim waiver is effective, the first sub-state of the claim waiver element is determined. Then, based on the first claim rejection answer in the claim rejection path and the logical connection between each claim rejection establishment element, as well as the first sub-state of each claim waiver element, the second sub-state of the claim rejection establishment element is determined. The second state includes both the first sub-state of the claim waiver element and the second sub-state of the claim rejection establishment element.

[0108] Scenario 6: The claim denial path includes multiple elements for establishing a claim denial and multiple elements for waiving a claim denial. It is understood that cases 4 and 5 are less complex than case 6. Therefore, this embodiment will describe case 6 in detail. Cases 4 and 5 can be found in the description of case 6.

[0109] In some embodiments, the elements of a claim rejection path include elements for establishing a claim rejection and elements for waiving a claim rejection; there are multiple elements for establishing a claim rejection, and these elements are logically connected; there are multiple elements for waiving a claim rejection, and these elements are logically connected; based on the first claim rejection answer of the claim rejection path, the second state of the elements of a claim rejection path is determined, including: Based on the first rejection answer in the rejection path, whether the rejection waiver is effective in the rejection path, and the logical connection relationship between the various rejection waiver elements in the rejection path, determine the first sub-state of each rejection waiver element in the rejection path; the first sub-state indicates whether the fact corresponding to the rejection waiver element is true; Based on the first rejection answer of the rejection path, the first sub-state of each rejection exemption element in the rejection path, and the logical connection relationship between each rejection establishment element in the rejection path, the second sub-state of each rejection establishment element in the rejection path is determined; the second sub-state indicates whether the fact corresponding to the rejection establishment element is true; wherein, the second state includes the first sub-state of the rejection exemption element and the second sub-state of the rejection establishment element.

[0110] The process can be described in Table 3 above. In this process, when there are multiple elements for claim waiver, the logical connections between each element constitute the claim waiver conditions. When there are multiple elements for claim acceptance, the logical connections between each element constitute the claim acceptance conditions. Meeting the claim waiver conditions does not mean that the claim waiver is effective; the prerequisite for the claim waiver to be effective is that both the claim waiver conditions and the claim acceptance conditions are met.

[0111] The total state value of each rejection exemption element in Table 3 can be considered as meeting the rejection exemption conditions, and the total state value of each rejection exemption element in Table 3 can be considered as not meeting the rejection exemption conditions. The total state value of each rejection establishment element in Table 3 can be considered as meeting the rejection establishment conditions, and the total state value of each rejection establishment element in Table 3 can be considered as not meeting the rejection establishment conditions.

[0112] During this process, if the claim waiver is effective in the claim rejection path, the total state value of all claim waiver elements in Table 3 needs to be 1. Based on this, and considering the logical connections between the various claim waiver elements, the first sub-state of each claim waiver element in the claim rejection path is determined. The logical connections include at least one of "AND" and "OR". If the claim waiver is not effective in the claim rejection path, the total state value of all claim waiver elements in Table 3 needs to be 0. Based on this, and considering the logical connections between the various claim waiver elements, the first sub-state of each claim waiver element in the claim rejection path is determined. The logical connections include at least one of "AND" and "OR". Next, based on the first claim rejection answer of the claim rejection path, the first sub-state of each claim waiver element in the claim rejection path, and the logical connections between the various claim rejection establishment elements in the claim rejection path, the second sub-state of each claim rejection establishment element in the claim rejection path is determined.

[0113] In one example, referring to Table 3, if the first rejection answer is no compensation, and the total state value of the rejection exemption element obtained from the first sub-state of each rejection exemption element is 0, then the total state value of the rejection establishment element obtained from the second sub-state value of each rejection establishment element is 1. Furthermore, based on the logical connection relationship between each rejection establishment element, the first sub-state of each rejection establishment element in the rejection path is determined.

[0114] For another example, if the first rejection answer is a claim, and the total state value of the rejection exemption elements obtained from the first sub-state of each rejection exemption element is 1, then the total state value of the rejection establishment elements obtained from the second sub-state value of each rejection establishment element is determined to be 1. Furthermore, based on the logical connection relationship between each rejection establishment element, the first sub-state of each rejection establishment element in the rejection path is determined. In this example, the rejection exemption takes effect.

[0115] Therefore, in this embodiment, for the above situation 6, the first sub-state of obtaining the elements of claim rejection exemption and the second sub-state of obtaining the elements of claim rejection can be accurately determined.

[0116] Figure 4 This specification provides a flowchart illustrating another data generation method based on an insurance contract, which can be applied to and executed by a server, and the server may include... Figure 1 Server 101 in the middle. For example... Figure 4 As shown, the method includes: Step S402: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content. Step S404: Determine the claim answer to the claim problem to be constructed; based on the claim answer, determine the first sub-claim answer of the claim unit for the claim problem and the second sub-claim answer of the claim rejection unit for the claim problem. Step S406: Based on the first sub-claim answer, determine the first claim answer for each claim path for the claim issue; based on the second sub-claim answer and the configuration information regarding whether the claim rejection waiver is effective in the claim answer, determine the first rejection answer for each rejection path for the claim issue. Step S408: For each claim path, determine the first state of the claim element in the claim path based on the first claim answer of the claim path. The first state indicates whether the fact corresponding to the claim element is true. For each denial path, determine the second state of the denial element in the denial path based on the first denial answer of the denial path. The second state indicates whether the fact corresponding to the denial element is true. Step S410: Based on the first state of the compensation element in each compensation path and the second state of the rejection element in each rejection path, construct the claims problem and obtain the problem construction process data of the claims problem; Step S412: Reason about the claims issue to obtain the problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, generate the claims question and answer data corresponding to the insurance contract based on the claims issue and the claims answer.

[0117] Figure 4 The specific process can be found in the previous description, and will not be repeated here.

[0118] Taking the aforementioned insurance contract A as an example, it has four payment paths: 1, 2, 1, and 2.

[0119] Claim path 1 includes: claim basis node W "Insured hospitalized due to accidental injury", claim condition node D " (Accident) (Hospitalization), payout node C "payment" and supporting node B "accidental injury refers to an external, non-disease-related sudden event"; Claim path 2 includes: claim basis node W "hospitalization allowance will be paid if hospitalization is caused by illness after the 30-day waiting period", and claim condition node D " (After a 30-day waiting period) (disease) (Hospitalization), payout node C node "Payout"; Claim denial path 1 includes: claim denial basis node W "No compensation for drunk driving", and compensation condition node D " (Drunk driving), claim denial node C "claim denial"; Pathway 2 for claim denial includes: node W, the basis for claim denial: "No compensation will be paid for cosmetic surgery, except for necessary facial repairs caused by accidental injury"; and node D, the establishment of claim denial. (Cosmetic Surgery), Node C "Rejection" and Node "Rejection Waiver" (Unexpected facial repair).

[0120] Through the above process, the first answer to claim payment path 1 is determined to be a claim, the first answer to claim payment path 2 is determined to be no claim, the first answer to claim rejection path 1 is determined to be a claim and the claim rejection waiver takes effect, and the first answer to claim rejection path 2 is also determined to be a claim.

[0121] Next, determine the compensation elements in compensation path 1. (Accident) The first state (hospitalization) has a state value of 1, indicating that the corresponding fact is established. Determine the compensation elements in compensation path 2. (After a 30-day waiting period) and The first state of (disease) is 0, indicating that the corresponding fact is not true, thus confirming... The first state value of (hospitalization) is 1, indicating that the corresponding fact is true. and They can be considered as the same compensation element.

[0122] Determine the elements for establishing a claim denial in Path 1 The second state (second sub-state) of (drunk driving) has a state value of 0, indicating that the corresponding fact is not established. Determine the elements for establishing a claim denial in claim denial path 2. The second state (second sub-state) of (plastic surgery) has a state value of 1, indicating that the corresponding fact is established. Determine the elements for claim exemption in claim rejection path 2. The state value of the second state (first sub-state) of (accidental facial repair) is 1, indicating that the corresponding fact is true.

[0123] Next, using a large language model, based on the first state of each compensation element, the second state (second sub-state) of the elements for establishing a claim denial, and the second state (first sub-state) of the elements for waiving a claim denial, the following claim question can be derived: I just bought this medical insurance 3 days ago. Yesterday, I unfortunately suffered a car accident resulting in severe facial injuries and was hospitalized. The doctor performed facial reconstruction surgery on me. The traffic police determined that I did not drink alcohol. Can I claim compensation? Through the above process, by determining the status of the elements for compensation, the elements for justification of claim rejection, and the elements for waiver of claim rejection, the corresponding claims questions can be generated. The status values ​​of the elements for compensation, the elements for justification of claim rejection, and the elements for waiver of claim rejection can be called a "truth value matrix".

[0124] Figure 5 This is a schematic diagram illustrating the generation of claims issues based on an argumentation model, provided for one or more embodiments of this specification. Figure 5 As shown, the process of determining the first sub-claim answer, the second sub-claim answer, the first payout answer, the first rejection answer, the first state, and the second state in this embodiment can be called a top-down truth assignment process. Through this process, firstly, a global truth value is assigned, which includes the first and second sub-claim answers; this global truth value assignment also determines the first and second sub-claim answers. Next, clause truth values ​​are assigned, which include the first payout answer and the first rejection answer; this clause truth value assignment also determines the first payout answer and the first rejection answer. Finally, propositional truth values ​​are assigned. A proposition refers to the atomic propositions obtained from the aforementioned decomposition, namely, the payout element, the rejection validity element, and the rejection exemption element (if any). The propositional truth values ​​include the first state and the second state; this propositional truth value assignment also determines the first and second states. Furthermore, based on the first state of the compensation elements in each compensation path and the second state (second sub-state) of the elements for establishing the denial of claims in each denial of claims path, and the second state (first sub-state) of the elements for waiving the denial of claims, the claims problem is obtained through LLM construction.

[0125] Furthermore, in this embodiment, it is also necessary to obtain the problem construction process data of the claims issue. In one example, an argumentation model can be obtained through a large language model. Based on the claims answer, the compensation elements in the compensation path, and the rejection elements in the rejection path, a claims question corresponding to the claims answer is constructed. The large language model can output the claims question and the problem construction process data. Additionally, by using the same or different large language models as described above, reasoning can be performed on the claims question. The large language model can obtain the problem reasoning process data and determine whether the problem construction process data and the problem reasoning process data meet the consistency requirements.

[0126] In one embodiment, obtaining the problem construction process data for the claims issue includes: Obtain the first state of the payout element in the payout path and the second state of the rejection element in the rejection path; the first state indicates whether the fact corresponding to the payout element is true; the second state indicates whether the fact corresponding to the rejection element is true. Based on the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path, the problem construction process data is obtained; The claims settlement issue is constructed based on the first state of the claims settlement element in the claims settlement path and the second state of the claims denial element in the claims denial path.

[0127] In this embodiment, by obtaining the first state of the compensation element determined in the above process and the second state of the rejection element in the rejection path, it can be understood that the claims issue is constructed based on the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path. Furthermore, based on the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path, the problem construction process data is obtained.

[0128] As previously explained, in one scenario, the elements for claim rejection include elements for establishing claim rejection, and the second state includes a second sub-state of the elements for establishing claim rejection. The second sub-state indicates whether the facts corresponding to the elements for establishing claim rejection are true. In this scenario, the claim rejection path includes a node for establishing claim rejection. In another scenario, the elements for claim rejection include elements for establishing claim rejection and elements for waiving claim rejection. The second state includes a second sub-state of the elements for establishing claim rejection and a first sub-state of the elements for waiving claim rejection. The second sub-state indicates whether the facts corresponding to the elements for establishing claim rejection are true, and the first sub-state indicates whether the facts corresponding to the elements for waiving claim rejection are true. In this scenario, the claim rejection path includes a node for establishing claim rejection and a node for waiving claim rejection. Therefore, the problem construction process data can include the second sub-state, as well as the first and second sub-states, thus providing corresponding problem construction process data for both cases with and without a claim waiver.

[0129] In some cases, the first state of the payout element in the payout path and the second state of the denial element in the denial path can be used as the problem construction data.

[0130] In other cases, there is at least one payout path, and each payout path constitutes a payout unit; there is at least one rejection path, and each rejection path constitutes a rejection unit; based on the first state of the payout element in the payout path and the second state of the rejection element in the rejection path, the problem construction process data is obtained, including: The first state of the claim elements in each claim payment path, the second state of the claim rejection elements in each claim rejection path, the first claim payment answer for the claim issue in each claim payment path, the first claim rejection answer for the claim issue in each claim rejection path, the first sub-claim answer for the claim issue in the claim payment unit, and the second sub-claim answer for the claim issue in the claim rejection unit are used as the problem construction process data; wherein, the first claim payment answer is determined based on the first sub-claim answer; the first claim rejection answer is determined based on the second sub-claim answer and the configuration information of whether the claim rejection waiver is effective in the claim answer; the first sub-claim answer and the second sub-claim answer are determined based on the claim answer.

[0131] In this scenario, the first compensation answer, the first rejection answer, the first sub-claim answer, and the second sub-claim answer are obtained. The first state, the second state, the first compensation answer, the first rejection answer, the first sub-claim answer, and the second sub-claim answer are all used as problem construction process data. This allows the problem construction process data to comprehensively reflect the construction process of the claim problem and improve the comprehensiveness of the problem construction process data.

[0132] After obtaining the problem construction process data, in step S206 above, reasoning is performed on the claims problem to obtain problem reasoning process data. In some embodiments, reasoning on the claims problem to obtain problem reasoning process data includes: Determine the fourth state of the payout element in the payout path represented by the claim issue, and determine the fifth state of the rejection element in the rejection path represented by the claim issue; the fourth state indicates whether the facts corresponding to the payout element are true; the fifth state indicates whether the facts corresponding to the rejection element are true. Based on the fourth state of the compensation element in the compensation path and the fifth state of the denial element in the denial path, the problem reasoning process data is obtained.

[0133] As described above, the claim rejection path may include elements for establishing a claim rejection, or it may include both elements for establishing a claim rejection and elements for waiver of claim. In claims processing, the issues may include facts corresponding to elements for establishing a claim rejection, or facts corresponding to both elements for establishing a claim rejection and elements for waiver of claim. Therefore, the fifth state includes two scenarios: 1. The fifth state includes the fourth sub-state of elements for establishing a claim rejection, indicating whether the facts corresponding to the elements for establishing a claim rejection are valid. 2. The fifth state includes the third sub-state of elements for waiver of claim and the fourth sub-state of elements for establishing a claim rejection, where the third sub-state indicates whether the facts corresponding to the elements for waiver of claim are valid, and the fourth sub-state indicates whether the facts corresponding to elements for establishing a claim rejection are valid.

[0134] In this embodiment, firstly, the fourth state of the payout element in the payout path represented by the claim problem is determined, and the fifth state of the rejection element in the rejection path represented by the claim problem is determined. Specifically, the claim problem can be input into a large language model, along with the atomic propositions contained in the claim problem. Atomic propositions refer to the atomic propositions obtained from the aforementioned decomposition, including payout elements, rejection validity elements, and rejection waiver elements (if any). The large language model first performs text analysis on the claim problem to obtain the state of each atomic proposition within the claim problem. This state indicates whether the fact represented by the atomic proposition is true; this state includes the fourth and fifth states. Of course, when inputting atomic propositions into the large language model, it is optional to input an indication message indicating whether the atomic proposition belongs to a payout element, a rejection validity element, or a rejection waiver element.

[0135] Taking the aforementioned claim question, "I just bought this medical insurance three days ago. Yesterday, I was unfortunately involved in a car accident that resulted in severe facial injuries, and I was hospitalized. The doctor performed facial plastic reconstruction surgery on me. The traffic police determined that I had not been drinking. Can I claim compensation?" as an example, this claim question and the atomic propositions "accident," "hospitalization," "after 30-day waiting period," "illness," "drunk driving," "plastic surgery," and "accidental facial repair" are input into the large language model.

[0136] The large language model analyzes claims issues and can determine the state of the atomic proposition "accident" as 1, "hospitalization" as 1, "after 30-day waiting period" as 0, "illness" as 0, "drunk driving" as 0, "plastic surgery" as 1, and "accidental facial repair" as 1. Here, 1 indicates that the corresponding fact is true, and 0 indicates that the corresponding fact is false.

[0137] Next, using a large language model, the reasoning process data is obtained based on the fourth state of the compensation element in the compensation path and the fifth state of the rejection element in the rejection path. In one case, the fourth state of the compensation element in the compensation path and the fifth state of the rejection element in the rejection path can be used as the reasoning process data.

[0138] Therefore, through this embodiment, based on the ability of semantic analysis, semantic analysis can be performed on the claims issue to determine the fourth state of the compensation element in the compensation path represented by the claims issue, and the fifth state of the rejection element in the rejection path represented by the claims issue. Based on the fourth state of the compensation element in the compensation path and the fifth state of the rejection element in the rejection path, the problem reasoning process data can be accurately obtained.

[0139] In one embodiment, the number of payout paths is at least one, and each payout path constitutes a payout unit; the number of rejection paths is at least one, and each rejection path constitutes a rejection unit; based on the fourth state of the payout element in the payout path and the fifth state of the rejection element in the rejection path, the problem reasoning process data is obtained, including: For each claim path, based on the fourth state of the claim elements in the claim path, determine the second claim answer for the claim problem. Based on the second claim answer for each claim path, determine the first sub-reasoning answer for the claim problem for the claim unit. For each claim denial path, based on the fifth state of the claim denial element in the claim denial path, determine the second claim denial answer for the claim issue of the claim denial path; based on the second claim denial answer of each claim denial path, determine the second sub-reasoning answer of the claim denial unit for the claim issue. The fourth state of the compensation element, the fifth state of the denial element, each second compensation answer, each second denial answer, the first sub-reasoning answer, and the second sub-reasoning answer are used as data for the problem reasoning process.

[0140] This needs to be discussed in two scenarios: 1. When inputting atomic propositions into the large language model, input an indication message indicating whether the atomic proposition belongs to a payout element, a grounds for denial of claim, or a grounds for denial of claim exemption. 2. When inputting atomic propositions into the large language model, do not input an indication message indicating whether the atomic proposition belongs to a payout element, a grounds for denial of claim, or a grounds for denial of claim exemption.

[0141] For scenario 1, the large language model can read each payout path and each rejection path in the argumentation model, thereby determining the payout path to which the payout element belongs, and the rejection path to which the rejection validity element and rejection exemption element belong. For scenario 2, the large language model can read each payout path and each rejection path in the argumentation model, and based on the content of the payout condition nodes in the payout path and the rejection condition nodes in the rejection path, determine whether the atomic proposition belongs to a payout element, a rejection validity element, or a rejection exemption element, and determine the payout path to which the payout element belongs, and the rejection path to which the rejection validity element and rejection exemption element belong.

[0142] Next, for each payout path, the large language model determines the second payout answer for the claim problem based on the fourth state of the payout element in the payout path. Based on the second payout answers for each payout path, it determines the first sub-reasoning answer for the claim problem for each payout unit. Furthermore, for each denial path, based on the fifth state of the denial element in the denial path, it determines the second denial answer for the claim problem for each denial path. Based on the second denial answer for each denial path, it determines the second sub-reasoning answer for the denial unit.

[0143] Finally, the large language model uses the fourth state of the payout element, the fifth state of the denial element, each second payout answer, each second denial answer, the first sub-reasoning answer, and the second sub-reasoning answer as data for the problem reasoning process.

[0144] It is understood that the data used in the problem construction process and the data used in the problem reasoning process need to be consistent. If the data used in the problem construction process includes the first state of the payout element in the payout path and the second state of the rejection element in the rejection path, then the data used in the problem reasoning process includes the fourth state of the payout element in the payout path and the fifth state of the rejection element in the rejection path. If the data used in the problem construction process includes the first state of the payout element in the payout path, the second state of the rejection element in the rejection path, the first payout answer, the first rejection answer, the first sub-claim answer, and the second sub-claim answer, then the data used in the problem reasoning process includes the fourth state of the payout element, the fifth state of the rejection element, the second payout answer, the second rejection answer, the first sub-reasoning answer, and the second sub-reasoning answer.

[0145] Therefore, through this embodiment, not only can the fourth state of the compensation element and the fifth state of the denial element be used as problem reasoning process data, but also the second compensation answer, the second denial answer, the first sub-reasoning answer, and the second sub-reasoning answer can be used as problem reasoning process data, making the problem reasoning process data more comprehensive and better able to reflect the complete reasoning process of the claims issue.

[0146] As mentioned earlier, each payout path may include one payout element or multiple payout elements.

[0147] In one scenario, the compensation path includes a single compensation element. In this case, for each compensation path, the second compensation answer for the claim issue can be determined based on the fourth state of the compensation element within the path. If the fourth state indicates that the fact corresponding to the compensation element is true, then the second compensation answer is determined to be compensation. If the fourth state indicates that the fact corresponding to the compensation element is false, then the second compensation answer is determined to be no compensation.

[0148] In another scenario, the compensation path includes multiple compensation elements, which are logically connected. In this case, for each compensation path, based on the fourth state of the compensation elements within the path, a second compensation answer for the claim problem is determined. This includes determining the second compensation answer for each compensation path based on the fourth state of the compensation elements and the logical connections between them.

[0149] In practice, you can refer to the process of determining the first state of the compensation elements when there are multiple compensation elements. The reverse process of this process is the process of determining the second compensation answer. Here, we will only use a few examples to illustrate this, and the specific process will not be repeated.

[0150] Example 1: The compensation path includes elements A, B, and C. In the compensation condition node, the relationship between elements A, B, and C is that all three have corresponding facts that are true; that is, they are connected by a logical AND. If the first states of elements A, B, and C all indicate that their corresponding facts are true, then the second compensation answer is determined to be compensation.

[0151] Example 2: The compensation path includes elements A, B, and C. In the compensation condition node, the relationship between elements A, B, and C is that any one of them corresponds to a fact that is true; that is, they are logically ORed. If the first state of at least one of elements A, B, and C indicates that the corresponding fact is true, then the second compensation answer is determined to be compensation.

[0152] Example 3: The various compensation elements in the compensation path include element A, element B, and element C, which can be represented as "(element A)" Element B) "Element C". If the first state of at least one of elements A and B indicates that the corresponding fact is true, and the first state of element C indicates that the corresponding fact is true, then the second compensation answer is determined to be compensation.

[0153] Therefore, through this embodiment, when the compensation path includes multiple compensation elements, the second compensation answer for each compensation path can be accurately determined based on the fourth state of the compensation elements in the compensation path and the logical connection relationship between each compensation element in the compensation path.

[0154] After determining the second payout answer for each payout path regarding the claim issue, a first sub-reasoning answer for the payout unit regarding the claim issue is also determined based on the second payout answer for each payout path. Specifically, if the second payout answer for at least one payout path is a payout, then the first sub-reasoning answer indicates a payout; if the second payout answer for each payout path is a non-payment, then the first sub-reasoning answer indicates a non-payment.

[0155] As mentioned above, each denial path includes one or more denial elements and may or may not include one or more denial exemption elements.

[0156] Scenario 1: The claim denial process includes one element that would justify a claim denial. In this scenario, for each denial path, the second denial answer can be determined based on the fifth state of the elements constituting the denial in that path. If the fifth state indicates that the facts corresponding to the elements constituting the denial are true, then the second denial answer is denial. If the fifth state indicates that the facts corresponding to the elements constituting the denial are false, then the second denial answer is payment.

[0157] Scenario 2: The claim denial process includes multiple elements for establishing a claim denial. In this case, for each denial path, a second denial answer can be determined based on the fifth state of the elements establishing denial in the denial path and the logical connections between these elements. This process can be referenced from determining the second payout answer when there are multiple payout elements, and will not be repeated here.

[0158] Scenario 3: The claim denial path includes one element for establishing a claim denial and one element for waiving a claim denial. In this case, you can refer to Table 3 above to determine the second denial answer. Specifically, substitute the fifth state of the elements for establishing a denial and the fifth state of the elements for waiving a denial into Table 3 above to obtain the second denial answer of either not paying or paying the claim.

[0159] Scenario 4: The claim denial path includes one element for establishing a claim denial and multiple elements for waiver of claim denial. In this case, it is necessary to obtain the corrected rejection answer for the rejection path based on the fifth state of each rejection exemption element and the logical connection relationship between them. The corrected answer is the total state value of multiple rejection exemption elements in Table 3, which indicates whether the rejection exemption conditions are met. A total state value of 1 indicates that the conditions are met, and a total state value of 0 indicates that the conditions are not met. Next, based on the fifth state of the rejection establishment element, the corrected rejection answer is modified to obtain the second rejection answer of either non-payment or payment. Modification means that if the rejection exemption conditions are met, regardless of whether the facts corresponding to the fifth state of the rejection establishment element are true or false, the second rejection answer is payment; if the rejection exemption conditions are not met, the facts corresponding to the fifth state of the rejection establishment element are true, and the second rejection answer is non-payment; if the rejection exemption conditions are not met, the facts corresponding to the fifth state of the rejection establishment element are false, and the second rejection answer is payment. This process can be referred to in Table 3.

[0160] Scenario 5: The claim denial path includes multiple elements for establishing a claim denial and one element for waiving a claim denial. In this case, based on the fifth state of the elements for claim rejection exemption, a revised rejection answer for the rejection path needs to be obtained. This revised answer corresponds to the total state value of multiple elements for claim rejection exemption in Table 3, indicating whether the rejection exemption conditions are met. When the facts corresponding to the elements for claim rejection exemption are true, the rejection exemption conditions are met. When the facts corresponding to the elements for claim rejection exemption are false, the rejection exemption conditions are not met. Next, based on the fifth state of each element for claim rejection and the logical connection between them, the revised rejection answer is modified to obtain a second rejection answer of either non-payment or payment. Modification refers to obtaining the total state value of multiple elements for claim rejection in Table 3 based on the fifth state of each element for claim rejection and the logical connection between them. This total state value indicates whether the rejection conditions are met. A total state value of 1 indicates satisfaction, and a total state value of 0 indicates non-satisfaction. If the conditions for waiver of claim are met, the second answer to the claim denial will be to pay out, regardless of whether the conditions for establishing a claim denial are met. If the conditions for waiver of claim are not met, but the conditions for establishing a claim are met, the second answer to the claim denial will be to not pay out. If the conditions for waiver of claim are not met, but the conditions for establishing a claim are not met, the second answer to the claim denial will be to pay out. This process can be referred to in Table 3.

[0161] Scenario 6: The claim denial path includes multiple elements for establishing a claim denial and multiple elements for waiving a claim denial. It is understood that cases 4 and 5 are less complex than case 6. Therefore, this embodiment will describe case 6 in detail. Cases 4 and 5 can be found in the description of case 6.

[0162] The elements for claim rejection in the rejection path include elements for establishing a claim rejection and elements for waiving a claim rejection. There are multiple elements for establishing a claim rejection, and these elements are logically connected. There are also multiple elements for waiving a claim rejection, and these elements are logically connected. The fifth state includes the third sub-state of the waiving element and the fourth sub-state of the establishing element. The third sub-state indicates whether the fact corresponding to the waiving element is true, and the fourth sub-state indicates whether the fact corresponding to the establishing element is true. For each rejection path, based on the fifth state of the elements for rejection in the rejection path, the second rejection answer for the claim issue is determined, including: Based on the third sub-state of the exemption element in the claim rejection path and the logical connection relationship between each exemption element in the claim rejection path, the answer to the claim rejection path to be corrected is obtained. Based on the fourth sub-state of the elements that establish the claim rejection in the claim rejection path and the logical connection relationship between each element that establishes the claim rejection in the claim rejection path, the claim rejection answer to be corrected is modified to obtain the second claim rejection answer of the claim rejection path.

[0163] The process can be described in Table 3 above. In this process, when there are multiple elements for claim rejection exemption, the logical connections between each element constitute the claim rejection exemption conditions. When there are multiple elements for claim rejection establishment, the logical connections between each element constitute the claim rejection establishment conditions.

[0164] When the total state value of each claim waiver element is determined to be 1 based on the third sub-state of each claim waiver element and the logical connection relationship between each claim waiver element, the claim waiver condition is satisfied. When the total state value of each claim waiver element is determined to be 0 based on the third sub-state of each claim waiver element and the logical connection relationship between each claim waiver element, the claim waiver condition is not satisfied.

[0165] When the total state value of each element for establishing a claim rejection is determined to be 1 based on the fourth sub-state of each element and the logical connection between them, the conditions for establishing a claim rejection are met. When the total state value of each element for establishing a claim rejection is determined to be 0 based on the fourth sub-state of each element and the logical connection between them, the conditions for establishing a claim rejection are not met.

[0166] First, based on the third sub-state of the waiver element in the claim rejection path and the logical connection relationship between the various waiver elements in the claim rejection path, the answer to be corrected for the claim rejection path is obtained. This answer corresponds to the total state value of the multiple waiver elements in Table 3, indicating whether the waiver condition is met. Next, based on the fourth sub-state of the claim acceptance element in the claim rejection path and the logical connection relationship between the various claim acceptance elements in the claim rejection path, the answer to be corrected is modified to obtain a second claim rejection answer of either no payment or payment. Modification refers to obtaining the total state value of the multiple claim acceptance elements in Table 3 based on the fourth sub-state of the claim acceptance element in the claim rejection path and the logical connection relationship between the various claim acceptance elements in the claim rejection path. This total state value indicates whether the claim acceptance condition is met. If the waiver condition is met, the second claim rejection answer is payment regardless of whether the claim acceptance condition is met. If the waiver condition is not met, but the claim acceptance condition is met, the second claim rejection answer is no payment. If the waiver condition is not met, but the claim acceptance condition is not met, the second claim rejection answer is payment. The process can be referenced in Table 3.

[0167] Therefore, this embodiment enables the generation of a second denial response for the denial path in various situations. It should be noted that each of the above status values ​​is represented by "1" to indicate that the corresponding fact is true, and "0" to indicate that the corresponding fact is false.

[0168] After determining the second denial answer for each denial path regarding the claim issue, a second sub-reasoning answer for the denial unit regarding the claim issue is also determined based on the second denial answers for each denial path. Specifically, if the second payment answer for at least one denial path is no payment, then the second sub-reasoning answer indicates no payment; if the second denial answer for each denial path is a claim, then the second sub-reasoning answer indicates a claim.

[0169] In step S206 above, the data from the problem construction process is compared with the data from the problem reasoning process. If they are the same, the claim problem is deemed qualified, and the claim problem and its answer are used as the claim question and answer data corresponding to the insurance contract. If they are different, the claim problem and its answer are discarded.

[0170] Figure 6 This specification provides a flowchart illustrating another data generation method based on an insurance contract, which can be applied to and executed by a server, and the server may include... Figure 1 Server 101 in the middle. For example... Figure 6 As shown, the method includes: Step S602: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content. Step S604: Determine the claim answer to the claim question to be constructed; based on the claim answer, the payout elements in the payout path, and the rejection elements in the rejection path, construct the claim question corresponding to the claim answer. Step S606: Obtain the problem construction process data for the claims issue; the problem construction process data includes the first state of the claims element in the claims payment path and the second state of the claims rejection element in the claims rejection path; Step S608: Reason about the claims issue to obtain the problem reasoning process data; the problem reasoning process data includes the fourth state of the compensation element in the compensation path represented by the claims issue and the fifth state of the rejection element in the rejection path; Step S610: Determine whether the data from the problem construction process and the data from the problem reasoning process meet the consistency requirements; Step S612: If satisfied, generate claim Q&A data corresponding to the insurance contract based on the claim questions and claims answers; If the conditions are not met in step S614, then discard the claim question and the claim answer.

[0171] Figure 6 The specific process can be found in the previous description, and will not be repeated here.

[0172] Using insurance contract A as an example, the following claims question arises: I only purchased this medical insurance three days ago. Yesterday, I was unfortunately involved in a car accident, resulting in severe facial injuries, and was hospitalized. Doctors performed facial reconstruction surgery on me. The traffic police determined that I had not been drinking. Can I claim compensation? The claims problem and the atomic propositions "accident," "hospitalization," "after 30-day waiting period," "illness," "drunk driving," "plastic surgery," and "accidental facial repair" are input into the large language model. The large language model reasones about the claims problem, determines whether the facts corresponding to each atomic proposition are true, and obtains the state of each atomic proposition. Based on the state of each atomic proposition, the second payout answer for each payout path and the second rejection answer for each rejection path are determined. Based on the second payout answer for each payout path, the first sub-reasoning answer of the payout unit for the claims problem is obtained. Based on the second rejection answer for each rejection path, the second sub-reasoning answer of the rejection unit for the claims problem is obtained.

[0173] The large language model compares the state of each atomic proposition with the first and second states mentioned above. If they are the same, the claim question and answer are used as the claim question and answer data corresponding to the insurance contract. Of course, the large language model can also obtain the second payout answer, the second denial answer, the first sub-reasoning answer, and the second sub-reasoning answer, and generate the final claim answer text, such as: "Claim accepted."

[0174] Figure 7 A schematic diagram illustrating the verification of claims issues based on an argumentation model, provided for one or more embodiments of this specification, such as... Figure 7 As shown, the process of determining the state of each atomic proposition, the second payout answer, the second rejection answer, the first sub-reasoning answer, and the second sub-reasoning answer in this embodiment can be called a bottom-up truth value determination process. Through this process, first, proposition truth value determination is performed, which includes the state of each atomic proposition; determining proposition truth value is also determining the state of each atomic proposition. Next, clause truth value determination is performed, which includes the second payout answer and the second rejection answer; determining clause truth value is also determining the second payout answer and the second rejection answer. Finally, global truth value determination is performed, which includes the first sub-reasoning answer and the second sub-reasoning answer; determining global truth value is also determining the first sub-reasoning answer and the second sub-reasoning answer. Furthermore, the large language model verifies whether the claim is eligible based on the state of each atomic proposition. Alternatively, it verifies whether the claim is eligible based on the state of each atomic proposition, the second payout answer, the second rejection answer, the first sub-reasoning answer, and the second sub-reasoning answer.

[0175] Figure 8 This diagram illustrates the data generation process of an insurance contract provided in one or more embodiments of this specification. Figure 8 It is equivalent to Figure 3 , Figure 5 , Figure 7 A schematic diagram of the integration process. Only the following is described here. Figure 8 and Figure 3 , Figure 5 , Figure 7 The differences. Figure 8 The middle part indicates Figure 3 , Figure 5 , Figure 7 The connection between the three attached figures. Figure 8 The text also illustrates the process of verifying claims issues, which requires comparing the results of truth value assignment with the results of truth value determination. Figure 8 The document also provides a step-by-step explanation of the process: obtaining the insurance contract, generating claims questions, verifying claims questions, and storing verified claims questions and answers in the database.

[0176] In summary, the above implementation methods for data generation achieve the following effects: 1. Through the systematic design of this embodiment, the shortcomings of related data construction schemes are overcome. The automated framework can generate thousands of high-quality data samples on a large scale and at low cost, which greatly improves efficiency. In addition, the controllable synthesis mechanism of this embodiment can generate complex edge cases that are few in number in the real world but are crucial to the evaluation of model capabilities, such as cases containing multiple layers of nested exclusion logic, as needed, thereby solving the problem of uneven distribution of real data. 2. This embodiment overcomes the fatal flaw of unreliable logic when simple large models suggest engineering synthesis data. This embodiment does not allow the model to "create" freely, but rather "fills in the blanks" within a strict logical framework. The introduction of the nested Toulmin model provides a solid skeleton for the reasoning process, the hierarchical truth assignment algorithm ensures that the generated content strictly serves the preset logical goal, and the final consistency verification stage can eliminate any claims issues with logical flaws. 3. This embodiment can stably generate a large-scale benchmark dataset (including claims questions and answers) specifically designed for insurance claims adjudication tasks. Each sample in this dataset possesses three excellent characteristics: First, logical consistency, ensuring no logical contradictions exist between the case description and the adjudication conclusion; second, complete interpretability, with each sample accompanied by a complete, step-by-step reasoning chain from atomic propositions to clause application and the final conclusion, providing material for training interpretable artificial intelligence models; and third, scenario diversity and realism, with the generated data achieving a high level of language fluency and professionalism, and covering a variety of complex logical scenarios. Utilizing such high-quality data, a deep analysis of the real logical reasoning capabilities of large-scale language models can be conducted, revealing their specific shortcomings in handling exclusion clauses and verifying missing conditions, thus pointing the way for subsequent model improvements.

[0177] Figure 9 This is a schematic diagram of the structure of a data generation device based on an insurance contract according to one or more embodiments of this specification, such as... Figure 9 As shown, the device includes: Model acquisition unit 92 acquires the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content; Problem construction unit 94 determines the claim answer to the claim problem to be constructed, constructs the claim problem corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and obtains the problem construction process data of the claim problem; The problem verification unit 96 performs reasoning on the claim problem to obtain problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, it generates claim question and answer data corresponding to the insurance contract based on the claim problem and the claim answer.

[0178] Optionally, the insurance contract includes one or more of the claim-related contents; it also includes a first creation unit, which, before obtaining the argumentation model corresponding to the insurance contract, creates a claim-related argumentation node for each claim-related content, and creates a claim-related path based on the claim-related argumentation node; wherein, the claim-related argumentation node includes: a claim-based node, a claim-condition node, and a claim-conclusion node; the claim-based node stores the claim-based content in the claim-related content; the claim-condition node stores the claim-related elements in the claim-related content; and the claim-conclusion node is used to represent the claim. Optionally, in the compensation condition node, the number of compensation elements is one; or, in the compensation condition node, the number of compensation elements is multiple, and the compensation elements have logical connection relationships.

[0179] Optionally, the insurance contract includes one or more of the aforementioned claim rejection-related content; it also includes a second creation unit, which, before obtaining the argumentation model corresponding to the insurance contract, creates a claim rejection argumentation node corresponding to each of the claim rejection-related content, and creates a claim rejection path corresponding to the claim rejection-related content based on the claim rejection argumentation node; wherein, the claim rejection argumentation node includes: a claim rejection basis node, a claim rejection condition node, and a claim rejection conclusion node; the claim rejection basis node stores the claim rejection basis content in the claim rejection-related content; the claim rejection condition node stores the claim rejection elements in the claim rejection-related content; and the claim rejection conclusion node is used to indicate a claim rejection.

[0180] Optionally, the claim rejection condition node includes a claim rejection establishment node, which stores the claim rejection establishment elements in the claim rejection related content; the claim rejection elements include the claim rejection establishment elements; in the claim rejection establishment node, the number of claim rejection establishment elements is one; or, in the claim rejection establishment node, the number of claim rejection establishment elements is multiple, and each claim rejection establishment element has a logical connection relationship.

[0181] Optionally, the claim rejection condition node includes a claim rejection establishment node and a claim rejection waiver node. The claim rejection establishment node stores the claim rejection establishment elements in the claim rejection-related content; the claim rejection waiver node stores the claim rejection waiver elements in the claim rejection-related content; the claim rejection elements include the claim rejection establishment elements and the claim rejection waiver elements; in the claim rejection establishment node, the number of claim rejection establishment elements is one or more, and when the number of claim rejection establishment elements is multiple, there is a logical connection relationship between each claim rejection establishment element; in the claim rejection waiver node, the number of claim rejection waiver elements is one or more, and when the number of claim rejection waiver elements is multiple, there is a logical connection relationship between each claim rejection waiver element.

[0182] Optionally, the number of claim paths is at least one, and each claim path constitutes a claim unit; the number of claim rejection paths is at least one, and each claim rejection path constitutes a claim rejection unit; the problem construction unit 94, based on the claim answer, determines a first sub-claim answer for the claim problem in the claim unit and a second sub-claim answer for the claim problem in the claim rejection unit; based on the first sub-claim answer, determines a first claim answer for the claim problem in each claim path; based on the second sub-claim answer and the configuration information regarding whether the claim rejection waiver is effective in the claim answer, determines a first claim rejection answer for the claim problem in each claim rejection path; and constructs the claim problem based on the first claim answer of each claim path, the first claim rejection answer of each claim rejection path, the claim elements in the claim path, and the claim rejection elements in the claim rejection path.

[0183] Optionally, the problem construction unit 94 obtains a first state of the compensation element in the compensation path and a second state of the rejection element in the rejection path; the first state indicates whether the fact corresponding to the compensation element is true; the second state indicates whether the fact corresponding to the rejection element is true; based on the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path, the problem construction process data is obtained; wherein, the claim problem is constructed based on the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path.

[0184] Optionally, the problem verification unit 96 determines the fourth state of the compensation element in the compensation path represented by the claim problem, and determines the fifth state of the rejection element in the rejection path represented by the claim problem; the fourth state indicates whether the fact corresponding to the compensation element is true; the fifth state indicates whether the fact corresponding to the rejection element is true; based on the fourth state of the compensation element in the compensation path and the fifth state of the rejection element in the rejection path, the problem reasoning process data is obtained.

[0185] Optionally, the number of the payout paths is at least one, and each payout path constitutes a payout unit; the number of the rejection paths is at least one, and each rejection path constitutes a rejection unit; the problem verification unit 96, for each payout path, determines a second payout answer for the claim problem based on the fourth state of the payout element in the payout path, and determines a first sub-reasoning answer for the claim problem based on the second payout answer of each payout path; for each rejection path, determines a second rejection answer for the claim problem based on the fifth state of the rejection element in the rejection path, and determines a second sub-reasoning answer for the claim problem based on the second rejection answer of each rejection path; the fourth state of the payout element, the fifth state of the rejection element, each second payout answer, each second rejection answer, the first sub-reasoning answer, and the second sub-reasoning answer are used as the problem reasoning process data.

[0186] For ease of description, the above devices are described by dividing them into various modules or units based on their functions. Of course, when implementing one or more of these specifications, the functions of each module or unit can be implemented in the same or different software and / or hardware, or a module that performs the same function can be implemented by a combination of multiple sub-modules or sub-units, etc. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0187] Since the apparatus embodiment corresponds to the method embodiment, the description is relatively simple. Figure 9 For a more detailed description of the apparatus, please refer to the description of the method embodiments, which will not be repeated here.

[0188] One or more embodiments of this specification also provide a data generation device based on an insurance contract, which is used to execute the data generation method based on an insurance contract provided above. Figure 10 This is a schematic diagram of the structure of a data generation device based on an insurance contract, provided for one or more embodiments of this specification.

[0189] like Figure 10 As shown, device 1000 mainly consists of a communication interface 1002, a user interface 1004, a processor 1006, and a data storage 1008. These components are interconnected and communicate with each other via a system bus, network, or other connection mechanism 810. The communication interface 1002 enables device 1000 to communicate with other devices, access networks, and transmission networks via analog or digital modulation. For example, the communication interface 1002 may include a chipset and antenna for wireless communication with a radio access network or access point. Furthermore, the communication interface 1002 can be a wired interface such as Ethernet, Token Ring, or a USB port, or a wireless interface such as Wi-Fi, Bluetooth, Global Positioning System (GPS), or a wide-area wireless interface (e.g., WiMAX or LTE). Of course, the communication interface 1002 can also support other forms of physical layer interfaces and standard or proprietary communication protocols. The communication interface 1002 may also include multiple physical communication interfaces, such as Wi-Fi, Bluetooth, and wide-area wireless interfaces.

[0190] User interface 1004 includes receiving user input and providing output to the user. Therefore, user interface 1004 may include input components such as a keypad, keyboard, touch-sensitive or presence-sensitive panel, computer mouse, trackball, joystick, microphone, still camera, and video camera, and output components such as a display screen (which may be combined with a touch-sensitive panel), CRT, LCD, LED, display using DLP technology, printer, and other similar devices known or developed in the future. User interface 1004 may also generate auditory output via speakers, speaker jacks, audio output ports, audio output devices, headphones, and other similar devices known or developed in the future. In some embodiments, user interface 1004 may include software, circuitry, or other forms of logic capable of transmitting and receiving data from external user input / output devices. Additionally or alternatively, device 1000 may support remote access from other devices via communication interface 1002 or another physical interface (not shown). User interface 1004 may be configured to receive user input, the position and movement of which may be indicated by indicators or cursors described herein. User interface 1004 can also be configured as a display device for rendering or displaying text fragments.

[0191] Processor 1006 may include one or more general-purpose processors and / or dedicated processors. Data storage 1008 may include one or more volatile and / or non-volatile storage components, and may be integrated wholly or partially with processor 1006. Data storage 1008 may include removable and non-removable components.

[0192] Processor 1006 is capable of executing program instructions 818 (e.g., compiled or uncompiled program logic and / or machine code) stored in data storage 1008 to perform the various functions described herein. Data storage 1008 may contain a non-transitory computer-readable medium on which program instructions are stored, which, when executed by device 1000, enable device 1000 to perform any methods, processes, or functions disclosed in this specification and / or the accompanying drawings. Execution of program instructions 818 by processor 1006 may result in processor 1006 using data 812. For example, program instructions 818 may include an operating system 822 (e.g., an operating system kernel, device drivers, and / or other modules) installed on device 1000 and one or more application programs 820 (e.g., a browser, social application, or game application).

[0193] Similarly, data 812 may contain operating system data 816 and application data 814. Operating system data 816 is primarily accessible to the operating system 822, while application data 814 is primarily accessible to one or more applications 820. Application data 814 may reside in a file system visible or hidden from the user of device 1000. Applications 820 may communicate with the operating system 822 through one or more application programming interfaces (APIs). These APIs facilitate applications 820 reading and / or writing application data 814, transmitting or receiving information via communication interface 1002, receiving or displaying information on user interface 1004, etc. In some terms, application 820 may be simply referred to as an "app". Furthermore, application 820 may be downloaded to device 1000 through one or more online app stores or app markets. However, applications may also be installed on device 1000 in other ways, such as through a web browser or a physical interface on device 1000 (e.g., a USB port).

[0194] In one specific embodiment, the insurance contract-based data generation device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for the insurance contract-based data generation device, and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content; Determine the claim answer to the claim question to be constructed, and based on the claim answer, the claim elements in the claim payment path, and the claim rejection elements in the claim rejection path, construct the claim question corresponding to the claim answer, and obtain the question construction process data of the claim question; The claim problem is reasoned to obtain the problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, the claim question and answer data corresponding to the insurance contract is generated based on the claim problem and the claim answer.

[0195] Since the device embodiment corresponds to the method embodiment, the description is relatively simple. For more information about the device, please refer to the description of the method embodiment, which will not be repeated here.

[0196] This specification provides an embodiment of a computer-readable storage medium as follows: Corresponding to the above-described method for generating data based on an insurance contract, and based on the same technical concept, one or more embodiments of this specification also provide a computer-readable storage medium.

[0197] The computer-readable storage medium provided in this embodiment is used to store computer-executable instructions, which, when executed, implement the following process: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content; Determine the claim answer to the claim question to be constructed, and based on the claim answer, the claim elements in the claim payment path, and the claim rejection elements in the claim rejection path, construct the claim question corresponding to the claim answer, and obtain the question construction process data of the claim question; The claim problem is reasoned to obtain the problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, the claim question and answer data corresponding to the insurance contract is generated based on the claim problem and the claim answer.

[0198] It should be noted that the embodiments of a computer-readable storage medium described in this specification and the embodiments of a data generation method based on an insurance contract described in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can be referred to the implementation of the corresponding method described above, and the repeated parts will not be described again.

[0199] This specification provides an example of a computer program product as follows: Corresponding to the above-described method for generating data based on an insurance contract, and based on the same technical concept, one or more embodiments of this specification also provide a computer program product.

[0200] A computer program product includes a computer program / instructions that, when executed by a processor, perform the following steps: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes compensation-related content and claim rejection-related content; the argumentation model includes the compensation path corresponding to the compensation-related content and the claim rejection path corresponding to the claim rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the claim rejection path stores the claim rejection elements in the claim rejection-related content; Determine the claim answer to the claim question to be constructed, and based on the claim answer, the claim elements in the claim payment path, and the claim rejection elements in the claim rejection path, construct the claim question corresponding to the claim answer, and obtain the question construction process data of the claim question; The claim problem is reasoned to obtain the problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, the claim question and answer data corresponding to the insurance contract is generated based on the claim problem and the claim answer.

[0201] It should be noted that the embodiments of a computer program product described in this specification and the embodiments of a data generation method based on an insurance contract described in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can be referred to the implementation of the corresponding method described above, and the repeated parts will not be described again.

[0202] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments. For example, the device embodiment, equipment embodiment and computer-readable storage medium embodiment are all similar to the method embodiment, so the description is relatively simple. When reading the relevant content of the device embodiment, equipment embodiment and computer-readable storage medium embodiment, please refer to the description of the method embodiment.

[0203] While one or more embodiments of this specification provide method steps as described in the embodiments or flowcharts, it is understood that the order of steps listed in the embodiments or flowcharts is merely one possible execution order among many steps, and does not represent the only execution order. Therefore, when the claims involve method steps, any changes or adjustments to the order of such steps, or the parallelism between steps, are also within the scope of protection of the claims. This specification uses specific terms to describe embodiments of this specification. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different locations in this specification do not necessarily refer to the same embodiment. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0204] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0205] In the 1930s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many improvements to the methodology today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that an improvement to the methodology cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must also be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0206] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0207] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0208] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0209] Those skilled in the art will understand that one or more embodiments of this specification can be provided as a method, system, or computer program product. Therefore, one or more embodiments of this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this specification may take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0210] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0211] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0212] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0213] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0214] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0215] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0216] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising at least one…" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0217] It should also be noted that the terms "one," "an," and "the" do not specifically refer to the singular; they can also include the plural. Ordinal numbers such as "first," "second," etc., do not necessarily indicate order; often they are used to distinguish objects. For example, "first server" and "second server" usually refer to two servers. To differentiate between these two servers, they are described as "first server" and "second server." Of course, sometimes these two servers may be the same server. Unless explicitly stated, "receiving and sending data" does not necessarily mean direct receipt and transmission; it can be indirect. For example, A receiving data sent by B can be understood as A directly receiving data sent by B, or it can be understood as A indirectly receiving data sent by B through other entities such as C. Similarly, B sending data to A can be understood as B sending data directly to A, or it can be understood as B indirectly sending data to A through other entities such as C. Here, C can be one entity, or it can be two or more entities.

[0218] Unless explicitly stated otherwise, the relationships between structures can be direct or indirect. For example, when describing "A is connected to B," unless it is explicitly stated that A and B are directly connected, it should be understood that A can be directly connected to B or indirectly connected to B. Similarly, when describing "A is above B," unless it is explicitly stated that A is directly above B (AB is adjacent and A is above B), it should be understood that A can be directly above B or indirectly above B (AB is separated by other elements, and A is above B). And so on.

[0219] One or more embodiments of this specification can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. One or more embodiments of this specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0220] The above description is merely an embodiment of this document and is not intended to limit the scope of this document. Various modifications and variations can be made to this document by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this document should be included within the scope of the claims of this document.

Claims

1. A data generation method based on insurance contracts, comprising: Obtain the argumentation model corresponding to the insurance contract; the insurance contract includes content related to claims settlement and content related to claims denial. The argumentation model includes a compensation path corresponding to the compensation-related content and a rejection path corresponding to the rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the rejection path stores the rejection elements in the rejection-related content. Determine the claim answer to the claim question to be constructed, and based on the claim answer, the claim elements in the claim payment path, and the claim rejection elements in the claim rejection path, construct the claim question corresponding to the claim answer, and obtain the question construction process data of the claim question; The claim problem is reasoned to obtain the problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, the claim question and answer data corresponding to the insurance contract is generated based on the claim problem and the claim answer.

2. The method according to claim 1, wherein the insurance contract includes one or more of the claims-related contents; Before obtaining the argumentation model corresponding to the insurance contract, the following is also included: For each of the aforementioned compensation-related contents, a compensation verification node corresponding to the compensation-related contents is created, and a compensation path corresponding to the compensation-related contents is created based on the compensation verification node; The compensation justification node includes: a compensation basis node, a compensation condition node, and a compensation conclusion node; the compensation basis node stores the compensation basis content in the compensation-related content; the compensation condition node stores the compensation elements in the compensation-related content; and the compensation conclusion node is used to indicate compensation.

3. The method according to claim 2, wherein in the compensation condition node, the number of compensation elements is one; or, in the compensation condition node, the number of compensation elements is multiple, and each compensation element has a logical connection relationship.

4. The method according to claim 1, wherein the insurance contract includes one or more of the aforementioned claim denial-related contents; Before obtaining the argumentation model corresponding to the insurance contract, the following is also included: For each of the aforementioned claim rejection-related contents, a claim rejection argument node is created corresponding to the claim rejection-related contents, and a claim rejection path is created based on the claim rejection argument node. The claim rejection argument node includes: a claim rejection basis node, a claim rejection condition node, and a claim rejection conclusion node; the claim rejection basis node stores the claim rejection basis content in the claim rejection related content; the claim rejection condition node stores the claim rejection elements in the claim rejection related content; and the claim rejection conclusion node is used to indicate claim rejection.

5. The method according to claim 4, wherein the claim rejection condition node includes a claim rejection establishment node, and the claim rejection establishment node stores the claim rejection establishment elements in the claim rejection related content; the claim rejection elements include the claim rejection establishment elements; In the node where a claim is established, there is one element for establishing a claim; or, in the node where a claim is established, there are multiple elements for establishing a claim, and these elements are logically connected.

6. The method according to claim 4, wherein the claim rejection condition node includes a claim rejection establishment node and a claim rejection waiver node, wherein the claim rejection establishment node stores the claim rejection establishment elements in the claim rejection related content; the claim rejection waiver node stores the claim rejection waiver elements in the claim rejection related content; and the claim rejection elements include the claim rejection establishment elements and the claim rejection waiver elements; In the claim rejection establishment node, the number of claim rejection establishment elements is one or more, and when the number of claim rejection establishment elements is multiple, there is a logical connection relationship between each claim rejection establishment element; in the claim rejection waiver node, the number of claim rejection waiver elements is one or more, and when the number of claim rejection waiver elements is multiple, there is a logical connection relationship between each claim rejection waiver element.

7. The method according to claim 1, wherein the number of the compensation path is at least one, and each of the compensation paths constitutes a compensation unit; the number of the rejection path is at least one, and each of the rejection path constitutes a rejection unit; the step of constructing the claim question corresponding to the claim answer based on the claim answer, the compensation element in the compensation path, and the rejection element in the rejection path includes: Based on the claims response, the first sub-claims response of the payout unit and the second sub-claims response of the denial unit for the claims issue are determined. Based on the first sub-claim answer, determine the first claim answer for each claim path for the claim issue; based on the second sub-claim answer and the configuration information regarding whether the claim rejection waiver is effective in the claim answer, determine the first rejection answer for each claim rejection path for the claim issue. The claims problem is constructed based on the first payout answer for each payout path, the first rejection answer for each rejection path, the payout element in the payout path, and the rejection element in the rejection path.

8. The method according to claim 7, wherein the claim answer includes the answer of the argumentation model for the claim problem; and determining, based on the claim answer, a first sub-claim answer of the compensation unit for the claim problem and a second sub-claim answer of the rejection unit for the claim problem includes: If the stated claim answer indicates a claim, then the first sub-claim answer is determined to indicate a claim, and the second sub-claim answer is determined to indicate a claim. If the claim answer indicates no compensation, then the first sub-claim answer is determined to be a claim and the second sub-claim answer is determined to be a no-compensation; or, the first sub-claim answer is determined to be a no-compensation and the second sub-claim answer is determined to be a no-compensation; or, the first sub-claim answer is determined to be a no-compensation and the second sub-claim answer is determined to be a claim.

9. The method according to claim 7, wherein determining the first rejection answer for each rejection path for the claim issue based on the configuration information of whether the rejection waiver is effective in the second sub-claim answer and the claim answer, comprises: If the configuration information indicates that the claim rejection waiver is effective in the claim answer, then the first claim rejection answer of the claim rejection path related to the claim rejection waiver indicates a claim, and, based on the second sub-claim answer, the first claim rejection answer of the claim rejection path unrelated to the claim rejection waiver is determined; If the configuration information indicates that the claim rejection waiver is not effective in the claim answer, then the first rejection answer for each of the rejection paths is determined based on the second sub-claim answer.

10. The method according to claim 7, wherein constructing the claim problem based on the first payout answer for each payout path, the first rejection answer for each rejection path, the payout element in the payout path, and the rejection element in the rejection path comprises: For each of the compensation paths, a first state of the compensation element in the compensation path is determined based on the first compensation answer of the compensation path; The first state indicates whether the facts corresponding to the compensation element are true; For each of the claim rejection paths, a second state of the claim rejection element in the claim rejection path is determined based on the first claim rejection answer of the claim rejection path; the second state indicates whether the fact corresponding to the claim rejection element is true. The claims problem is constructed based on the first state of the claims element in each claims payment path and the second state of the claims rejection element in each claims rejection path.

11. The method according to claim 10, wherein the number of compensation elements in the compensation path is multiple, and each compensation element has a logical connection relationship; determining the first state of the compensation element in the compensation path based on the first compensation answer of the compensation path includes: Based on the first compensation answer of the compensation path and the logical connection relationship between each compensation element in the compensation path, the first state of each compensation element in the compensation path is determined.

12. The method according to claim 10, wherein the elements of claim rejection in the claim rejection path include elements for establishing claim rejection and elements for waiving claim rejection; the number of elements for establishing claim rejection is multiple, and each element for establishing claim rejection has a logical connection relationship; the number of elements for waiving claim rejection is multiple, and each element for waiving claim rejection has a logical connection relationship. The step of determining the second state of the rejection element in the rejection path based on the first rejection answer of the rejection path includes: Based on the first rejection answer of the rejection path, whether the rejection waiver is effective in the rejection path, and the logical connection relationship between each rejection waiver element in the rejection path, the first sub-state of each rejection waiver element in the rejection path is determined; the first sub-state indicates whether the fact corresponding to the rejection waiver element is true. Based on the first rejection answer of the rejection path, the first sub-state of each rejection exemption element in the rejection path, and the logical connection relationship between each rejection establishment element in the rejection path, the second sub-state of each rejection establishment element in the rejection path is determined; the second sub-state indicates whether the fact corresponding to the rejection establishment element is true; the second state includes the first sub-state and the second sub-state.

13. The method according to claim 1, wherein obtaining the problem construction process data of the claims issue includes: Obtain the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path; The first state indicates whether the facts corresponding to the claim element are true; the second state indicates whether the facts corresponding to the claim rejection element are true. Based on the first state of the compensation element in the compensation path and the second state of the rejection element in the rejection path, the problem construction process data is obtained; The claims issue is constructed based on the first state of the claims element in the claims payment path and the second state of the claims rejection element in the claims rejection path.

14. The method according to claim 13, wherein the number of claim payment paths is at least one, and each claim payment path constitutes a claim payment unit; the number of claim rejection paths is at least one, and each claim rejection path constitutes a claim rejection unit; obtaining the problem construction process data based on the first state of the claim payment element in the claim payment path and the second state of the claim rejection element in the claim rejection path includes: The first state of the claim element in each claim payment path, the second state of the claim rejection element in each claim rejection path, the first claim payment answer for the claim problem in each claim payment path, the first claim rejection answer for the claim problem in each claim rejection path, the first sub-claim answer for the claim problem in each claim payment unit, and the second sub-claim answer for the claim problem in the claim rejection unit are used as the problem construction process data; Wherein, the first compensation answer is determined based on the first sub-claim answer; the first claim rejection answer is determined based on the second sub-claim answer and the configuration information of whether the claim rejection waiver is effective in the claim answer; the first sub-claim answer and the second sub-claim answer are determined based on the claim answer.

15. The method according to claim 13, wherein the elements for claim rejection include elements for establishing claim rejection; the second state includes a second sub-state of the elements for establishing claim rejection; the second sub-state indicates whether the fact corresponding to the elements for establishing claim rejection is established; or, The elements for claim rejection include elements for establishing claim rejection and elements for waiving claim rejection; the second state includes a second sub-state of the elements for establishing claim rejection and a first sub-state of the elements for waiving claim rejection; the second sub-state indicates whether the facts corresponding to the elements for establishing claim rejection are true; the first sub-state indicates whether the facts corresponding to the elements for waiving claim rejection are true.

16. The method according to claim 1, wherein reasoning about the claims issue to obtain the reasoning process data includes: The fourth state of the payout element in the payout path represented by the claim issue is determined, and the fifth state of the rejection element in the rejection path represented by the claim issue is determined; the fourth state indicates whether the fact corresponding to the payout element is true; the fifth state indicates whether the fact corresponding to the rejection element is true. The problem reasoning process data is obtained based on the fourth state of the compensation element in the compensation path and the fifth state of the rejection element in the rejection path.

17. The method according to claim 16, wherein the number of the compensation path is at least one, and each of the compensation paths constitutes a compensation unit; the number of the rejection path is at least one, and each of the rejection path constitutes a rejection unit; the step of obtaining the problem reasoning process data based on the fourth state of the compensation element in the compensation path and the fifth state of the rejection element in the rejection path includes: For each of the claim paths, based on the fourth state of the claim elements in the claim path, a second claim answer for the claim problem is determined for the claim path; based on the second claim answer for each of the claim paths, a first sub-reasoning answer for the claim problem is determined for the claim unit. For each of the claim denial paths, based on the fifth state of the claim denial element in the claim denial path, a second claim denial answer for the claim issue is determined for the claim denial path; based on the second claim denial answer for each of the claim denial paths, a second sub-reasoning answer for the claim issue is determined for the claim denial unit. The fourth state of the compensation element, the fifth state of the denial element, each of the second compensation answers, each of the second denial answers, the first sub-reasoning answer, and the second sub-reasoning answer are used as the reasoning process data for the problem.

18. The method according to claim 17, wherein the number of compensation elements in the compensation path is multiple, and each compensation element has a logical connection relationship; determining the second compensation answer of the compensation path for the claim issue based on the fourth state of the compensation elements in the compensation path includes: Based on the fourth state of the compensation element in the compensation path and the logical connection relationship between each compensation element in the compensation path, a second compensation answer for the claim problem is determined for each compensation path.

19. The method according to claim 17, wherein the elements of claim rejection in the claim rejection path include elements for establishing claim rejection and elements for waiving claim rejection; the number of elements for establishing claim rejection is multiple, and each element for establishing claim rejection has a logical connection relationship; the number of elements for waiving claim rejection is multiple, and each element for waiving claim rejection has a logical connection relationship; the fifth state includes a third sub-state of the element for waiving claim rejection and a fourth sub-state of the element for establishing claim rejection; the third sub-state indicates whether the fact corresponding to the element for waiving claim rejection is true; the fourth sub-state indicates whether the fact corresponding to the element for establishing claim rejection is true; determining the second claim rejection answer of the claim rejection path for the claim issue based on the fifth state of the elements of claim rejection in the claim rejection path includes: Based on the third sub-state of the exclusion element in the exclusion path and the logical connection relationship between each exclusion element in the exclusion path, the exclusion answer to be corrected in the exclusion path is obtained; Based on the fourth sub-state of the elements establishing the claim rejection in the claim rejection path and the logical connection relationship between each element establishing the claim rejection in the claim rejection path, the claim rejection answer to be corrected is corrected to obtain the second claim rejection answer of the claim rejection path.

20. A data generation device based on an insurance contract, comprising: The model acquisition unit acquires the argumentation model corresponding to the insurance contract; the insurance contract includes content related to compensation and content related to claim rejection; The argumentation model includes a compensation path corresponding to the compensation-related content and a rejection path corresponding to the rejection-related content; the compensation path stores the compensation elements in the compensation-related content; the rejection path stores the rejection elements in the rejection-related content. The problem construction unit determines the claim answer to the claim problem to be constructed, constructs the claim problem corresponding to the claim answer based on the claim answer, the claim elements in the claim path, and the claim rejection elements in the claim rejection path, and obtains the problem construction process data of the claim problem; The problem verification unit performs reasoning on the claim problem to obtain problem reasoning process data. If the problem construction process data and the problem reasoning process data meet the consistency requirements, it generates claim question and answer data corresponding to the insurance contract based on the claim problem and the claim answer.

Images