Insurer support system

Abstract

To provide an insurer support system for presenting a statistical estimation result of whether a claim is right or wrong.SOLUTION: And an AI device 300 for statistically estimating whether or not the claim of the damage insurance money by the estimation sheet information is right and outputting the estimation result when the estimation sheet information necessary for calculating the damage insurance money amount is input, and insurer terminals 500 and for transmitting the estimation sheet information necessary for calculating the damage insurance money amount to the AI device 300 and receiving the estimation result output from the device.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an insurance provider support system, and more particularly to an insurance provider support system that presents statistical estimates of the success or failure of a property insurance claim. [Background technology]

[0002] Patent Document 1 discloses a program for estimating the amount of damage caused by a disaster at low cost and in a short time. This program causes a computer to function as a means for acquiring the results of a measurement of visible information carried out on buildings damaged by a disaster after the disaster occurs, and as a means for estimating the amount of damage to the damaged buildings by applying an estimation model constructed using machine learning about the relationship between the appearance of multiple buildings damaged by past disasters after the disaster occurs and the actual amount of damage to input data based on the results of the measurement of visible information. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2023-162107 Summary of the Invention [Problem to be solved by the invention]

[0004] In recent years, fraudulent claims involving unscrupulous insurance claim support companies and repair companies have become more prevalent as natural disasters become more frequent and severe, and the methods used are said to be becoming more sophisticated.Fraudulent claims for insurance are against social justice and undermine the foundations and trust of the insurance system.Allowing fraudulent claims could lead to a rise in insurance premiums, which could ultimately harm the interests of honest policyholders.

[0005] However, it can be said that eliminating fraudulent claims is difficult, since if it were easy to eliminate them, they would have been eradicated by now. This type of situation can occur not only with disaster insurance and earthquake insurance, which are claims made when natural disasters occur, but also with various types of non-life insurance, such as automobile insurance.

[0006] Therefore, the objective of this invention is to provide an insurance company support system that presents statistically estimated results on the success or failure of a property insurance claim. [Means for solving the problem]

[0007] In order to solve the above problems, the insurance provider support system of the present invention comprises: an artificial intelligence device that has learned estimate information related to fraudulent claims for non-life insurance as learning data, and when estimate information necessary for calculating the amount of non-life insurance is input, statistically estimates whether the claim for non-life insurance based on the estimate information is correct or not, and outputs the estimation result; an insurance company terminal that transmits estimate information necessary for calculating the non-life insurance amount to the artificial intelligence device and receives the estimation result output from the artificial intelligence device; Equipped with.

[0008] The artificial intelligence device also learns list information containing information on unscrupulous businesses as learning data, and the estimate information required to calculate the damage insurance amount may also include business information.

[0009] The artificial intelligence device can also output the estimation result together with the reason for the estimation.

[0010] Furthermore, a claim agent terminal for transmitting estimate information required for calculating the amount of the insurance premium to the insurance agent terminal may be provided. [Brief explanation of the drawings]

[0011] [Figure 1]FIG. 1 is a diagram illustrating an outline of an insurance provider support system according to an embodiment of the present invention, showing estimate information that a veteran employee of a non-life insurance company has determined to be highly likely to be a fraudulent claim. [Figure 2] FIG. 2 is a diagram showing quotation information different from the example in FIG. 1. [Figure 3] FIG. 3 is a diagram showing quotation information different from the examples of FIGS. 1 and 2. [Figure 4] 1 is a schematic configuration diagram of an insurance provider support system according to an embodiment of the present invention. [Figure 5] 5 is a block diagram showing the main components of a billing agent terminal 100, an insurance agent terminal 500, and a server device 200 in the insurance agent support system shown in FIG. 4. FIG. [Figure 6] 5 is a flowchart showing the operation of the insurance provider support system shown in FIG. 4. [Explanation of symbols]

[0012] 100 billing company terminals 110 Means of communication 120 Presentation means 130 Processing means 200 Server device 210 Means of communication 220 Processing means 230 Learning Tools 300 Artificial Intelligence Devices (hereinafter referred to as "AI devices") 400 Database (hereinafter referred to as "DB") 500 Insurance Company Terminals 600 Network DETAILED DESCRIPTION OF THE INVENTION

[0013] An insurance provider support system according to an embodiment of the present invention will be described below with reference to the accompanying drawings.

[0014] (Summary) For example, when claiming insurance for building construction work damaged by a natural disaster, the estimates provided by unscrupulous insurance claim support companies and repair contractors often have several things in common, such as the following:

[0015] 1. There are more billing items than necessary 2. The ratio of expenses to construction costs is high 3. The ratio of auxiliary construction costs to main construction costs is high 4. The basis for billing for material costs, etc. is not clearly stated.

[0016] Figures 1 to 3 show estimate information that a veteran insurance company employee judged to be highly likely to be a fraudulent claim. Please note that each estimate information has been slightly distorted, as presenting the actual estimate information as is could encourage fraudulent claims. Also, since each estimate information shown in Figures 1 to 3 was created by a different billing company, the format is not uniform.

[0017] The estimate information shown in Figure 1 is roughly divided into "Construction Cost (110)" totaling "775,000" yen and "Expenses (120)" totaling "290,000" yen. "Construction costs (110)" include the main construction cost of "Roof tile repair costs (111)" of 625,000 yen and the sub-construction cost of "Scaffolding assembly and dismantling costs (112)" of 150,000 yen. "Expenses (120)" includes "Related Expenses (121)" totaling "90,000" yen, "Related Disposal Expenses (122)" totaling "100,000" yen, and "Ancillary Work Expenses (123)" totaling "100,000" yen. For example, "Related Expenses (121)" includes "Miscellaneous Expenses (121a)" of 20,000 yen, "Management Expenses (121b)" of 30,000 yen, and "Site Management Expenses (121c)" of 40,000 yen.

[0018] The estimate information shown in Figure 1 corresponds to common points 1 to 4, namely, "1. There are more invoice items than necessary," "2. The ratio of expenses to construction costs is high," and "4. The basis for invoices for material costs, etc. is not clearly stated." The reasons for this are that, for common point 1, the total number of expense items alone is 11, for common point 1, the expense ratio is approximately 37.4% (= 290,000 yen / 775,000 yen), and for common point 4, the unit price of the roof tiles is unknown.

[0019] The estimate information shown in Figure 2 is also roughly divided into "Construction Costs (210)" totaling 310,000 yen and "Expenses (220)" totaling 557,000 yen. "Construction costs (210)" include, for example, "temporary construction costs (211)" which are sub-construction costs of 55,000 yen, "soil removal construction costs (212)" which are sub-construction costs of 60,000 yen, and "tile roofing costs (213)" which are main construction costs of 65,000 yen. "Expenses (220)" includes, for example, "waste disposal expenses (221)" of 185,000 yen, "safety measures expenses (222)" of 155,000 yen, and "miscellaneous work, scaffolding construction, and protective sheeting expenses (223)" of 80,000 yen.

[0020] The estimate information shown in Figure 2 corresponds to common points 1 to 4, namely, "1. There are more billing items than necessary," "2. The ratio of expenses to construction costs is high," and "4. The basis for billing for material costs, etc. is not clearly stated." The reasons for this are that, for common point 1, the total of construction items and expense items is 10, and for common point 1, the expense ratio is approximately 180% (= 557,000 yen / 310,000 yen).

[0021] In addition, the estimate information shown in Figure 2 may be double-counted, with the "cost of laying protective sheeting" included in "Miscellaneous work, scaffolding assembly, and protective sheeting costs (223)" and the "cost of protective work" included in "cost of moving, removing, and laying protective work (225)."

[0022] The estimate information shown in Figure 3 is also roughly divided into "Construction Costs (310)" totaling "448,000" yen and "Expenses (320)" totaling "18,000" yen. "Construction cost (310)" includes, for example, a sub-construction cost of "298,000" yen called "Temporary construction cost (311)" and a main construction cost of "140,000" yen called "Repair construction cost (312)". For example, "Temporary construction costs (311)" includes "◆Scaffolding construction costs (311a)" of "180,000" yen, and "◆Scaffolding-related construction costs (311b)" including "Road use permission procedure costs (311ba)" of "38,000" yen and "Guard personnel costs (311bb)" of "80,000" yen. "Expenses (320)" includes "◆Industrial waste expenses (321)."

[0023] The estimate information shown in Figure 3 corresponds to "3. The ratio of auxiliary construction costs to main construction costs is high" among common points 1 to 4. The reason for this is that the ratio is approximately 207% (= 290,000 yen / 140,000 yen).

[0024] In particular, the quotation information shown in Figure 3 is a relatively high and rare item, and the so-called unusual "security guard labor costs (311bb)" may not be appropriate, so this should be confirmed.

[0025] For example, if an AI device is trained using estimate information that a veteran employee has judged to be highly likely to be a fraudulent claim, such as the estimate information shown in Figures 1 to 3, and then the estimate information to be judged is input into the AI ​​device, if the estimate information contains the common points 1 to 4 above, is suspected of double counting as explained using Figure 2, or contains an unusual item such as the ``security guard labor costs'' as explained using Figure 3, an output can be obtained indicating that there is a statistically high possibility that a property insurance claim based on that estimate information is a fraudulent claim.

[0026] (Configuration explanation) Figure 4 is a schematic diagram of an insurance provider support system according to an embodiment of the present invention. Figure 4 shows a billing agent terminal 100, a server device 200, an AI device 300, a DB 400, an insurance provider terminal 500, and a network 600, which will be described below. Note that although Figure 4 shows only one of each terminal and device, in reality, multiple devices exist.

[0027] The billing agent terminal 100 is a terminal operated by a billing agent such as a property insurance claim support company or a repair company. The specific functions of the billing agent terminal 100 will be described later using Figure 5, but it mainly transmits estimate information for repair work costs and on-site photos that support the contents to the insurance agent terminal 500, and receives the examination results based on the estimation results by the AI ​​device 300 transmitted from the insurance agent terminal 500.

[0028] A typical example of the results of such an investigation would include a notice that if the claim is deemed to be legitimate, the damage insurance amount will be paid in accordance with the estimate information, and if the claim is deemed to be fraudulent, a notice that the estimate information needs to be revised. Furthermore, if the claim is not deemed to be fraudulent but a re-examination is required, a notice that further on-site photographs that substantiate the extent of the damage, etc., are required.

[0029] Furthermore, the billing agent terminal 100 may be, for example, a notebook personal computer, but is not limited to this. However, it is also possible to use information processing devices such as desktop personal computers, tablet terminals, mobile phones, smartphones, and PDAs (Personal Digital Assistants).

[0030] The server device 200 is managed by an administrator of the insurance provider support system of this embodiment. The server device 200 may be, for example, a business computer that is capable of wireless or wired communication with the AI ​​device 300 and the insurance provider terminal 500 via the network 600. Specific functions of the server device 200 will be described later with reference to FIG. 5.

[0031] The AI ​​device 300 is a device that utilizes a mathematical model such as an artificial intelligence, such as an artificial neural network. The AI ​​device 300 is trained to determine whether or not the estimate information sent from the claimant terminal 100 via the insurance company terminal 500 is an attempt to make a fraudulent claim prepared by an unscrupulous claimant, i.e., to statistically estimate whether or not the claim for the damage insurance amount is correct, and to output the estimation result.

[0032] DB400 is a storage medium that stores a list of unscrupulous businesses, which includes the names, email addresses, and other contact information of billing businesses, unique billing business IDs, and corporate information of unscrupulous businesses, as well as pairs of estimate information and corresponding estimation results that constitute learning data for continuously training AI device 300. The list of unscrupulous businesses includes the names, addresses, telephone numbers, email addresses, and URLs assigned to the websites of unscrupulous businesses.

[0033] For example, unscrupulous businesses may change only the name of their company or trade name without changing the address and / or telephone number, or may change the name but only partially (for example, changing the "construction" part of the name "ABC Construction Co., Ltd." to "ABC Repair Service Co., Ltd."), so it is a good idea to also use a list of unscrupulous businesses to determine whether or not there have been any fraudulent claims.

[0034] The list of unscrupulous businesses may be one that is already shared among multiple non-life insurance companies, or may be one that is separately created by an administrator of the insurance provider support system of this embodiment.

[0035] Furthermore, the quotation information constituting the learning data referred to here can typically be the quotation information shown in Figures 1 to 3 that an experienced staff member has judged to be highly likely to be a fraudulent claim in the early stages of operation of the insurance provider support system, but after that it can be estimated quotation information that the AI ​​device 300 has judged to be highly likely to be a fraudulent claim.

[0036] That is, in this embodiment, the AI ​​device 300 is adapted to deal with increasingly sophisticated fraudulent claims by having the server device 200 continuously learn using the learning data. The timing of this learning may be, for example, each time the number of learning data stored in the DB 400 reaches a certain number, or each time a predetermined time has elapsed since the previous learning.

[0037] The insurance company terminal 500 is a terminal operated by an insurance company employee, etc. The specific functions of the insurance company terminal 500 will be described later using Fig. 5, but the insurance company terminal 500 mainly receives estimate information for repair work costs transmitted from the billing agent terminal 100 and transmits the estimate information to the server device 200, receives estimation results transmitted from the server device 200, and transmits examination results based on the estimation results by the AI ​​device 300.

[0038] The insurance company terminal 500 may be, for example, a notebook personal computer, but is not limited to this. However, it is also possible to use information processing devices such as a desktop personal computer, a tablet terminal, a mobile phone, a smartphone, and a PDA.

[0039] Network 600 interconnects the billing agent terminal 100, server device 200, AI device 300, and DB 400, and is a general term for various networks including the Internet, an intranet that connects several of these locally, or a mobile phone network.

[0040] For example, Figure 4 shows an example in which the AI ​​device 300 and DB 400 are provided separately from the server device 200, but at least one of these may be built into the server device 200, as long as the required devices are present in the entire system.

[0041] Fig. 5 is a block diagram showing the main components of the billing agent terminal 100, the insurance company terminal 500, and the server device 200 in the insurance company support system shown in Fig. 4. Fig. 5(a) shows a block diagram of the billing agent terminal 100, Fig. 5(b) shows a block diagram of the insurance company terminal 500, and Fig. 5(c) shows a block diagram of the server device 200.

[0042] As shown in FIG. 5(a), the billing agent terminal 100 comprises a communication means 110, a presentation means 120, and a processing means 130, which will be described below.

[0043] The communication means 110 transmits and receives various information and data to and from other devices shown in Fig. 4. For example, the communication means 110 transmits estimate information for repair work costs and on-site photos that support the contents to the insurance company terminal 500 via the network 600 using functions normally provided in a web browser or mailer, and receives the examination results based on the estimation results by the AI ​​device 300 transmitted from the insurance company terminal 500.

[0044] Although the estimate information may be prepared in an unstructured format as shown in Figures 1 to 3, in this embodiment, the insurance company prepares a standardized estimate web page including input or selection fields for the estimate information, and the claim agent creates the estimate information through the web page. In this way, the comparison items between the estimate information are clarified, improving the estimation accuracy by the AI ​​device 300.

[0045] The presentation means 120 presents various information and data to the operator of the billing agent terminal 100. For example, the presentation means 120 outputs image signals to a display attached to the billing agent terminal 100 via a web browser installed in the billing agent terminal 100, such as a web page for estimates and a web page for examination results, which will be described below. The presentation means 120 can also present the information on these web pages by voice through a speaker. This also applies to the presentation means 520 of the insurance company terminal 500.

[0046] In this embodiment, the insurance company prepares a web page for the examination results that includes a display column for the examination results, and the presentation means 120 can present the calculation results of the damage insurance amount, requests to correct the estimate information, requests to submit further site photographs, etc. to the claim agent through the web page.

[0047] The processing means 130 performs various processes in the billing agent terminal 100. For example, the processing means 130 adds a unique billing agent ID previously assigned by the insurance company to the estimate information created by inputting or selecting through the estimate web page before the estimate information is sent by the communication means 110. The unique billing agent ID is used as information indicating the sender.

[0048] As shown in FIG. 5(b), the insurance company terminal 500 includes a communication means 510, a presentation means 520, and a processing means 530, which will be described below.

[0049] The communication means 510 transmits and receives various information and data to and from other devices shown in Fig. 4. For example, by using functions normally provided in a web browser or mailer, the communication means 510 receives estimate information for repair work costs transmitted from the billing agent terminal 100 via the network 600 and transmits the estimate information to the server device 200, as well as receiving estimation results transmitted from the server device 200 and transmitting review results based on the estimation results by the AI ​​device 300.

[0050] The presentation means 520 presents various information and data to the operator of the insurance company terminal 500. The presentation means 520 outputs image signals to a display attached to the insurance company terminal 500 via a web browser installed in the insurance company terminal 500, for example, to display a web page containing information such as estimate information on repair work costs from the billing company terminal 100 received by the communication means 510, estimation results from the server device 200, and calculation results of the damage insurance amount transmitted by the communication means 510.

[0051] The processing means 530 performs various processes in the insurance company terminal 500. For example, the processing means 530 performs various processes in the insurance company terminal 500, such as adding an insurance company ID used as information indicating the sender to a calculation of the insurance amount or a request to correct the insurance amount transmitted by the communication means 510.

[0052] As shown in FIG. 5(c), the server device 200 includes a communication unit 210, a processing unit 220, and a learning unit 230, which will be described below.

[0053] The communication means 210 transmits and receives various information and data to and from other devices, etc., shown in Fig. 4. For example, the communication means 210 receives estimate information from the billing agent terminal 100, which is transmitted from the insurance company terminal 500, and transmits to the insurance company terminal 500 the estimation result that is output when the estimate information is input to the AI ​​device 300.

[0054] The processing means 220 performs various processes in the server device 200. For example, the processing means 220 stores a pair of quotation information received by the communication means 210 and the corresponding estimation result in the DB 400 as learning data, or inputs the quotation information into the AI ​​device 300 and obtains the estimation result from the AI ​​device 300.

[0055] The learning means 230 continuously trains the AI ​​device 300 using the learning data stored in the DB 400. The learning means 230 also trains the AI ​​device 300 using the list information of unscrupulous businesses stored in the DB 400 as learning data.

[0056] (Explanation of operation) Figure 6 is a flow chart showing the operation of the insurance provider support system shown in Figure 4. The operation of the insurance provider support system shown in Figure 4 will be explained using Figure 6. It is assumed that the AI ​​device 300 has completed learning in the initial stage of operation using the estimate information shown in Figures 1 to 3.

[0057] When a disaster occurs and the claimant, who is the operator of the claimant terminal 100, wishes to claim insurance for the damages, the claimant creates estimate information by entering the details and amount of the insurance claim related to the repair work costs into an estimate web page that includes an input field for the estimate content, which the claimant receives by accessing the insurance company's homepage, etc.

[0058] Next, in accordance with the operation of the billing agent, the billing agent's terminal 100 imports on-site photographs that support the extent of damage, etc., in the estimate information, and the communication means 110 transmits the estimate information and the on-site photographs to the insurance company's terminal 500 via the network 600 (step S1).

[0059] At the insurance company terminal 500, when the communication means 510 receives the estimate information and the site photographs, the communication means 510 transmits the estimate information to the server device 200 via the network 600 in accordance with the operation of the insurance company (step S2).

[0060] In the server device 200, when the communication means 210 receives the quotation information, the processing means 220 outputs the quotation information to the AI ​​device 300 that has completed predetermined learning (step S3).

[0061] When the AI ​​device 300 inputs the estimate information, it statistically estimates whether the claim for the damage insurance amount is correct or not based on the contents of the estimate information, and outputs the estimation result to the server device 200.At this time, it is advisable to attach the reason for the estimation to the estimation result, as this may help the insurance company make a decision (step S4).

[0062] In the server device 200, when the communication means 210 inputs the estimation result, the communication means 210 transmits the estimation result to the insurance company terminal 500 via the network 600 (step S5), and the processing means 220 outputs a pair of the quotation information and the estimation result to the DB 400 as learning data (step S6).

[0063] In the insurance company terminal 500, when the communication means 510 receives the estimation result, the presentation means 520 presents the estimation result to the insurance company, thereby making it possible to support the insurance company in determining whether or not the claim for damage insurance based on the estimate information and the on-site photograph corresponding to the estimation result is likely to be fraudulent.

[0064] In the insurance company terminal 500, the communication means 510 will transmit the examination results based on the estimation results by the AI ​​device 300 in accordance with the operation of the insurance company.If the insurance company's judgment result is valid, it will send a message via the network 600 requesting that the damage insurance amount be paid in accordance with the estimate information, or if it is invalid, a message requesting that the estimate information be corrected (step S7).

[0065] Thereafter, in the insurance company terminal 500, for example, every time a predetermined time has elapsed since the previous learning, the learning means 230 continuously learns the AI ​​device 300 using the learning data stored in the DB 400 (step S8).

[0066] In addition, when the review result requesting that the quotation information be revised is sent in step S7, the insurance company may also include the reason for the AI ​​device 300's estimation, to the extent that presenting it does not encourage fraudulent claims.

[0067] As described above, the insurance provider support system of this embodiment inputs the estimate information necessary for calculating the amount of non-life insurance into an artificial intelligence device that has been trained using estimate information related to fraudulent claims as learning data, and statistically estimates the validity of the claim based on the estimate information, and can present the estimation result. This not only reduces the workload of insurance providers when a non-life insurance claim is made, but also enables them to present objective examination results to claimants.

Claims

1. an artificial intelligence device that has learned estimate information related to fraudulent claims for non-life insurance as learning data, and when estimate information necessary for calculating the amount of non-life insurance is input, statistically estimates whether the claim for non-life insurance based on the estimate information will be successful or not, and outputs the estimation result; an insurance company terminal that transmits estimate information necessary for calculating the non-life insurance amount to the artificial intelligence device and receives the estimation result output from the artificial intelligence device; An insurance provider support system comprising:

2. The artificial intelligence device has learned list information containing information on unscrupulous businesses as learning data, and the quote information required for calculating the amount of non-life insurance also includes information on the businesses.

2. The insurance provider support system according to claim 1.

3. The artificial intelligence device outputs the estimation result together with the reason for the estimation.

2. The insurance provider support system according to claim 1.

4. Further, a claim agent terminal is provided for transmitting estimate information necessary for calculating the non-life insurance amount to the insurance agent terminal.

2. The insurance provider support system according to claim 1.

Images