Car insurance anti-fraud method and device, electronic equipment and storage medium
By training machine learning models to detect vehicle accident data and auto insurance policy data, the problem of inefficient existing auto insurance anti-fraud methods is solved, and more efficient auto insurance anti-fraud detection is achieved.
Patent Information
- Application Number
- CN202411252983.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2024-12-03
AI Technical Summary
The existing anti-fraud methods of auto insurance are inefficient, mainly due to the low efficiency of manual inspection, which leads to the corresponding reduction in the anti-fraud detection efficiency of auto insurance.
By obtaining vehicle accident data samples and auto insurance policy data samples, training preset machine learning models, obtaining auto insurance anti-fraud models, and using this model to detect accident site data and target auto insurance data, reducing the need for manual inspection.
It improves the efficiency of anti-fraud detection of auto insurance, reduces the consumption of human resources, and achieves faster and more accurate anti-fraud detection of auto insurance.
Smart Images

Figure CN119067788A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence and financial technology, and in particular to a method and device for anti-fraud in auto insurance, an electronic device and a storage medium. Background Art
[0002] In insurance claims scenarios in the financial field, it is usually necessary to detect relevant accident data in order to achieve the purpose of insurance anti-fraud. For example, in the claims scenario of auto insurance, it is necessary to detect the accident data of the accident vehicle (such as accident location, accident time, accident speed, etc.) in order to achieve anti-fraud detection of auto insurance.
[0003] At present, the common anti-fraud method for auto insurance is that surveyors go to the accident site of the vehicle to conduct an investigation, and then detect whether it is a real vehicle accident based on the accident scene data obtained through manual investigation. Due to the low efficiency of manual investigation, the efficiency of auto insurance anti-fraud is low. Therefore, how to improve the efficiency of auto insurance anti-fraud has become a technical problem that needs to be solved urgently. Summary of the invention
[0004] The main purpose of the embodiments of the present application is to propose a car insurance anti-fraud method and device, an electronic device and a storage medium, aiming to improve the efficiency of car insurance anti-fraud.
[0005] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a method for anti-fraud of automobile insurance, which is applied to a service end, and the method comprises:
[0006] Obtain vehicle accident data samples and vehicle insurance policy data samples;
[0007] Based on the vehicle accident data sample and the auto insurance policy data sample, a preset machine learning model is trained to obtain an auto insurance anti-fraud model;
[0008] In response to a vehicle insurance report request sent by a client and carrying accident scene data, a target accident vehicle is determined based on the accident scene data;
[0009] Obtaining the vehicle insurance data of the target accident vehicle from a preset vehicle insurance database to obtain target vehicle insurance data;
[0010] Based on the auto insurance anti-fraud model, auto insurance anti-fraud detection is performed on the accident scene data and the target auto insurance data.
[0011] In some embodiments, the auto insurance anti-fraud model includes an accident data detection sub-model and an auto insurance data detection sub-model, and the auto insurance anti-fraud detection is performed on the accident scene data and the target auto insurance data based on the auto insurance anti-fraud model, including:
[0012] Based on the accident data detection sub-model, the accident scene data is subjected to a motor vehicle insurance anti-fraud detection to obtain an accident data detection result, wherein the accident data detection result is used to indicate whether the accident scene data is normal or abnormal;
[0013] Based on the vehicle insurance data detection sub-model, performing vehicle insurance anti-fraud detection on the target vehicle insurance data to obtain a vehicle insurance data detection result, wherein the vehicle insurance data detection result is used to indicate whether the target vehicle insurance data is normal or abnormal;
[0014] Based on the vehicle insurance data detection result, the accident data detection result is verified to obtain an accident data verification result, wherein the accident data detection result is used to characterize whether the accident scene data is normal or abnormal.
[0015] In some embodiments, the performing of auto insurance anti-fraud detection on the accident scene data based on the accident data detection sub-model to obtain the accident data detection result includes:
[0016] Converting the accident scene data into a format to obtain accident scene format data;
[0017] Extracting features from the accident scene format data to obtain accident scene feature data;
[0018] Based on the accident data detection sub-model and the gradient boosting decision tree algorithm, the accuracy of the accident scene feature data is estimated to obtain the accident data detection result.
[0019] In some embodiments, extracting features from the accident scene format data to obtain accident scene feature data includes:
[0020] Performing time extraction on the accident scene format data to obtain accident occurrence time data;
[0021] Extracting geographic information from the accident scene format data to obtain the accident geographic coordinates;
[0022] Extracting the vehicle speed from the accident scene format data to obtain the vehicle driving speed;
[0023] Extracting vehicle information from the accident scene format data to obtain a vehicle driving mode;
[0024] Extracting traffic road condition information from the accident scene format data to obtain traffic conditions;
[0025] The accident occurrence time data, the accident geographic coordinates, the vehicle driving speed, the vehicle driving mode and the traffic conditions are combined to obtain the accident scene characteristic data.
[0026] In some embodiments, the auto insurance data detection sub-model includes a factor decomposition machine and a deep neural network, and the auto insurance anti-fraud detection is performed on the target auto insurance data based on the auto insurance data detection sub-model to obtain the auto insurance data detection result, including:
[0027] Based on the vehicle insurance data detection sub-model, feature extraction is performed on the target vehicle insurance data to obtain vehicle insurance feature data;
[0028] Based on the factor decomposition machine, performing low-order feature interaction on the auto insurance feature data to obtain low-order feature interaction data;
[0029] Based on the deep neural network, high-order feature interaction is performed on the vehicle insurance feature data to obtain high-order feature interaction data;
[0030] The vehicle insurance data detection result is determined based on the low-order feature interaction data and the high-order feature interaction data.
[0031] In some embodiments, the verifying the accident data detection result based on the vehicle insurance data detection result to obtain the accident data verification result includes:
[0032] If the accident data detection result indicates that the accident scene data is normal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is normal, then generating investigation decision data, and determining the accident data verification result based on the investigation decision data;
[0033] If the accident data detection result indicates that the accident scene data is abnormal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is abnormal, generating the accident data verification result indicating that the accident scene data is abnormal;
[0034] If the accident data detection result indicates that the accident scene data is abnormal, or the vehicle insurance data detection result indicates that the target vehicle insurance data is abnormal, the investigation decision data is generated, and the accident data verification result is determined based on the investigation decision data.
[0035] In some embodiments, the vehicle insurance database includes a vehicle information field and a vehicle insurance data field, and obtaining the vehicle insurance data of the target accident vehicle from the preset vehicle insurance database to obtain the target vehicle insurance data includes:
[0036] Acquiring vehicle information of the target accident vehicle;
[0037] Based on the vehicle information, filtering out a target vehicle field from the vehicle information field;
[0038] The vehicle insurance data field of the target vehicle field in the vehicle insurance database is queried to obtain the target vehicle insurance data.
[0039] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a vehicle insurance anti-fraud device, the device comprising:
[0040] A training data acquisition module is used to obtain vehicle accident data samples and vehicle insurance policy data samples;
[0041] A machine model training module, used to train a preset machine learning model based on the vehicle accident data sample and the motor insurance policy data sample to obtain a motor insurance anti-fraud model;
[0042] An accident vehicle confirmation module, used to respond to a car insurance report request with accident scene data sent by a client, and determine a target accident vehicle based on the accident scene data;
[0043] An insurance data acquisition module is used to acquire the insurance data of the target accident vehicle from a preset insurance database to obtain target insurance data;
[0044] The auto insurance anti-fraud detection module is used to perform auto insurance anti-fraud detection on the accident scene data and the insurance data based on the auto insurance anti-fraud model.
[0045] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in the first aspect is implemented.
[0046] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.
[0047] The auto insurance anti-fraud method and device, electronic device and storage medium proposed in the present application use vehicle accident data samples and auto insurance policy data samples to train a preset machine learning model to obtain an auto insurance anti-fraud model, and when receiving an auto insurance report request carrying accident scene data sent by a client, use the above-mentioned auto insurance anti-fraud model to perform anti-fraud detection on the accident scene data and the target auto insurance data of the target accident vehicle, which can reduce the need for manual investigation, reduce the consumption of human resources, and improve the efficiency of auto insurance anti-fraud detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 is a flow chart of the auto insurance anti-fraud method provided in an embodiment of the present application;
[0049] Figure 2 yes Figure 1Flow chart of step S104 in FIG.
[0050] Figure 3 yes Figure 1 Flow chart of step S105 in FIG.
[0051] Figure 4 yes Figure 3 Flow chart of step S301 in FIG.
[0052] Figure 5 yes Figure 4 Flow chart of step S402 in FIG.
[0053] Figure 6 yes Figure 3 Flow chart of step S302 in FIG.
[0054] Figure 7 yes Figure 3 Flow chart of step S303 in FIG.
[0055] Figure 8 It is a structural schematic diagram of the auto insurance anti-fraud device provided in an embodiment of the present application;
[0056] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION
[0057] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0058] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.
[0060] First, some nouns involved in this application are analyzed:
[0061] Car insurance: Car insurance is an insurance product that provides protection for cars and motor vehicles. The main purpose of car insurance is to protect the financial interests of the car owner in the event of unexpected situations such as accidents, losses or theft.
[0062] Server: Server is a term in computer networking and software architecture, usually used to describe a computer or program responsible for providing resources, services, or data in a client-server model.
[0063] Client: "Client" is a term in computer networking and software architecture, usually used to describe the computer or program that interacts directly with the user and initiates requests in the client-server model.
[0064] In insurance claims scenarios in the financial field, it is usually necessary to detect relevant accident data in order to achieve the purpose of insurance anti-fraud. For example, in the claims scenario of auto insurance, it is necessary to detect the accident data of the accident vehicle (such as accident location, accident time, accident speed, etc.) in order to achieve anti-fraud detection of auto insurance.
[0065] At present, the common anti-fraud method for auto insurance is that surveyors go to the accident site of the vehicle to conduct an investigation, and then detect whether it is a real vehicle accident based on the accident scene data obtained through manual investigation. Due to the low efficiency of manual investigation, the efficiency of auto insurance anti-fraud is low. Therefore, how to improve the efficiency of auto insurance anti-fraud has become a technical problem that needs to be solved urgently.
[0066] Based on this, the embodiments of the present application provide a car insurance anti-fraud method and device, an electronic device and a storage medium, aiming to improve the efficiency of car insurance anti-fraud detection.
[0067] The auto insurance anti-fraud method and device, electronic device and storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the auto insurance anti-fraud method in the embodiments of the present application is described.
[0068] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0069] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0070] The auto insurance anti-fraud method provided in the embodiment of the present application relates to the field of artificial intelligence and financial technology. The auto insurance anti-fraud method provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the auto insurance anti-fraud method, etc., but is not limited to the above forms.
[0071] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.
[0072] Figure 1 is an optional flow chart of the auto insurance anti-fraud method provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S105.
[0073] Step S101, obtaining vehicle accident data samples and vehicle insurance policy data samples;
[0074] Step S102, training a preset machine learning model based on the vehicle accident data samples and the auto insurance policy data samples to obtain an auto insurance anti-fraud model;
[0075] Step S103, in response to the vehicle insurance report request carrying the accident scene data sent by the client, determining the target accident vehicle based on the accident scene data;
[0076] Step S104, obtaining the vehicle insurance data of the target accident vehicle from a preset vehicle insurance database to obtain the target vehicle insurance data;
[0077] Step S105, based on the auto insurance anti-fraud model, perform auto insurance anti-fraud detection on the accident scene data and the target auto insurance data.
[0078] In the steps S101 to S105 shown in the embodiment of the present application, the server trains the pre-built machine learning model according to the acquired vehicle accident data samples and the vehicle insurance policy data samples to obtain the vehicle insurance anti-fraud model for vehicle insurance anti-fraud detection, and upon receiving the vehicle insurance report request containing the accident scene data sent by the client, the target accident vehicle where the accident occurred is determined according to the accident scene data, and then the vehicle insurance data of the above target accident vehicle is obtained from the pre-built vehicle insurance database, and further, the authenticity of the vehicle insurance report request is detected using the above trained vehicle insurance anti-fraud model, the above accident scene data and the above target vehicle insurance data. Therefore, the present application obtains the vehicle insurance anti-fraud model by training the preset machine learning model using the vehicle accident data samples and the vehicle insurance policy data samples, and upon receiving the vehicle insurance report request carrying the accident scene data sent by the client, the above vehicle insurance anti-fraud model is used to perform anti-fraud detection on the accident scene data and the target vehicle insurance data of the target accident vehicle, which can reduce the need for manual investigation, reduce the consumption of human resources, and improve the efficiency of vehicle insurance anti-fraud detection.
[0079] In step S101 of some embodiments, the vehicle accident data sample may be vehicle accident data marked with whether there is insurance fraud, wherein the vehicle accident information may include information such as the location of the vehicle accident and the time of the vehicle accident. The vehicle insurance policy data sample may be vehicle insurance policy data marked with whether there is insurance fraud, wherein the vehicle insurance policy data may include information such as the total policy compensation amount and the vehicle damage compensation amount.
[0080] The present application can label the vehicle accident data of past vehicle accident events according to the accident results of manual investigation to obtain an initial sample of vehicle accident data, label the motor vehicle insurance policy data of past vehicle accident events according to the accident results of manual investigation to obtain an initial sample of motor vehicle insurance policy data, and then perform sample cleaning and data filling on the initial sample of vehicle accident data and the initial sample of motor vehicle insurance policy data to obtain a vehicle accident data sample and a motor vehicle insurance policy data sample.
[0081] In one application scenario, taking a vehicle collision accident as an example, based on the results of the manual investigation of the accident, the data such as the degree of vehicle collision loss, the speed of the accident vehicle, the geographical coordinates of the vehicle accident, the time of the vehicle accident, the driving mode of the accident vehicle, the traffic conditions at the accident site and the weather conditions at the accident site in the vehicle collision accident are labeled one by one to obtain a vehicle accident data sample, and then based on the results of the manual investigation of the accident, the policy data of the motor vehicle insurance purchased by the accident vehicle in the vehicle collision accident are labeled one by one to obtain a motor vehicle insurance policy data sample.
[0082] In step S102 of some embodiments, after obtaining vehicle accident data samples and vehicle insurance data samples, a pre-built machine learning model is trained based on the vehicle accident data samples and vehicle insurance data samples, and an accurate vehicle insurance anti-fraud model for vehicle insurance anti-fraud detection can be obtained, thereby reducing manual inspections of vehicle accident sites and improving the efficiency of vehicle insurance anti-fraud detection.
[0083] It is important to know that dividing the machine learning model into two sub-models, training them using vehicle accident data samples and vehicle insurance data samples respectively, and then combining the output results of the two sub-models to obtain the final output of the auto insurance anti-fraud model can effectively improve the accuracy of the auto insurance anti-fraud model.
[0084] Furthermore, by increasing the types of vehicle accident data samples, the accuracy of the auto insurance anti-fraud model can be improved. For example, existing vehicle accident data samples usually only contain the geographic coordinates of the vehicle accident, while the vehicle accident data samples in this application include data such as the speed of the accident vehicle, the geographic coordinates of the vehicle accident, the time when the vehicle owner reported the accident, the driving mode of the accident vehicle, and the traffic conditions at the accident location.
[0085] In step S103 of some embodiments, the client may be the driver of the vehicle in the accident or the purchaser of the auto insurance. The accident scene data may be information of the vehicle accident scene, and the accident scene data includes data such as the time of the accident, the geographic coordinates of the accident, the vehicle driving speed, the vehicle driving mode, the traffic conditions, the vehicle information, and the reporter information. The auto insurance report request may be an auto insurance reimbursement request including the vehicle accident event information sent by the client to the server. The target accident vehicle may be a vehicle that purchased auto insurance and had an accident at the vehicle accident scene.
[0086] In this application, when the server receives a car insurance report request containing accident scene data sent by the client, it can screen the accident scene data and extract the vehicle information of the vehicle involved in the accident.
[0087] In step S104 of some embodiments, the car insurance database may be pre-built in the server for storing vehicle information and car insurance data of vehicles that have purchased car insurance, and the car insurance database includes a vehicle information field and a car insurance data field, wherein the vehicle information field is used to store vehicle information, and the car insurance data field is used to store car insurance data. The car insurance data may be data such as the total policy compensation amount, the vehicle damage compensation amount, etc.
[0088] The present application can obtain target vehicle insurance data of the target accident vehicle by searching the vehicle insurance data of the target accident vehicle in the vehicle insurance database.
[0089] For details, see Figure 2 In some embodiments, step S104 may include but is not limited to steps S201 to S203:
[0090] Step S201, obtaining vehicle information of the target accident vehicle;
[0091] Step S202, based on the vehicle information, filter out the target vehicle field from the vehicle information field;
[0092] Step S203, query the vehicle insurance data field of the target vehicle field in the vehicle insurance database to obtain the target vehicle insurance data.
[0093] In step S201 of some embodiments, the vehicle information may be information used to represent the characteristics of the vehicle, such as the vehicle license plate number, vehicle model, and other information.
[0094] The present application obtains vehicle information by extracting information from the target accident vehicle. For example, when the target accident vehicle is image data, the license plate number of the target accident vehicle can be obtained by performing target recognition on the license plate in the image data, and then the vehicle model of the target accident vehicle can be obtained by performing target recognition on the vehicle body structure in the image data.
[0095] In step S202 of some embodiments, after the vehicle information of the target accident vehicle is determined, the target vehicle field of the target accident vehicle is found by traversing the vehicle information fields in the vehicle insurance database.
[0096] In step S203 of some embodiments, after the target vehicle field of the target accident vehicle is determined, the vehicle insurance data field corresponding to the target vehicle field is searched in the vehicle insurance database, and the vehicle insurance data in the found vehicle insurance data field is used as the target vehicle insurance data of the target accident vehicle.
[0097] In steps S201 to S203 shown in the present embodiment, the target vehicle insurance data of the target accident vehicle is obtained by searching the target vehicle field of the target accident vehicle in the pre-built vehicle insurance database and then searching the vehicle insurance data field corresponding to the target vehicle field in the vehicle insurance database, thereby improving the accuracy of the vehicle insurance data, reducing the time for querying the vehicle insurance data, and improving the efficiency of obtaining the vehicle insurance data.
[0098] In step S105 of some embodiments, after obtaining the accident scene data and target auto insurance data of the target accident vehicle, the accident scene data and the target auto insurance data are input into the above-mentioned trained auto insurance anti-fraud model for anti-fraud detection, which can detect whether the auto insurance report request sent by the client is consistent with the actual situation, thereby determining whether there is insurance fraud in this auto insurance report request.
[0099] It is important to know that in order to improve the detection accuracy of the auto insurance anti-fraud model, the auto insurance anti-fraud model includes an accident data detection sub-model and an auto insurance data detection sub-model. The accident data detection sub-model is used to detect whether there are abnormalities in the accident scene data, and the auto insurance data detection sub-model is used to detect whether there are abnormalities in the target auto insurance data.
[0100] Specifically, see Figure 3 In some embodiments, step S105 may include but is not limited to steps S301 to S303:
[0101] Step S301, based on the accident data detection sub-model, perform auto insurance anti-fraud detection on the accident scene data to obtain the accident data detection result, wherein the accident data detection result is used to indicate whether the accident scene data is normal or abnormal;
[0102] Step S302, based on the auto insurance data detection sub-model, perform auto insurance anti-fraud detection on the target auto insurance data to obtain the auto insurance data detection result, wherein the auto insurance data detection result is used to indicate whether the target auto insurance data is normal or abnormal;
[0103] Step S303, based on the vehicle insurance data detection result, the accident data detection result is verified to obtain an accident data verification result, wherein the accident data detection result is used to characterize whether the accident scene data is normal or abnormal.
[0104] In step S301 of some embodiments, the accident scene data is input into the accident data detection sub-model, which can implement auto insurance anti-fraud detection on the accident scene data, thereby obtaining an accident data detection result indicating whether the accident scene data is normal or abnormal.
[0105] See also Figure 4 In some embodiments, step S301 may include but is not limited to steps S401 to S403:
[0106] Step S401, converting the accident scene data into a format to obtain accident scene format data;
[0107] Step S402, extracting features from the accident scene format data to obtain accident scene feature data;
[0108] Step S403, based on the accident data detection sub-model and the gradient boosting decision tree algorithm, the accuracy of the accident scene feature data is estimated to obtain the accident data detection result.
[0109] In step S401 of some embodiments, to ensure that the accident scene data conforms to the input format of the accident data detection sub-model, the accident scene data may be format converted to obtain accident scene format data before being input into the accident data detection sub-model. For example, July 30 in the time data may be converted into Tuesday.
[0110] In step S402 of some embodiments, after the accident scene format data is obtained, feature extraction is performed on the accident scene format data to obtain the accident scene feature data required for performing anti-fraud detection of auto insurance using the accident data detection sub-model.
[0111] See also Figure 5 In some embodiments, step S402 may include but is not limited to steps S501 to S506:
[0112] Step S501, extracting time from the accident scene format data to obtain accident occurrence time data;
[0113] Step S502, extracting geographic information from the accident scene format data to obtain the accident geographic coordinates;
[0114] Step S503, extracting the vehicle speed from the accident scene format data to obtain the vehicle driving speed;
[0115] Step S504, extracting vehicle information from the accident scene format data to obtain a vehicle driving mode;
[0116] Step S505, extracting traffic information from the accident scene format data to obtain traffic conditions;
[0117] Step S506, combining the accident occurrence time data, the accident geographic coordinates, the vehicle driving speed, the vehicle driving mode and the traffic conditions to obtain the accident scene characteristic data.
[0118] In steps S501 to S505 of some embodiments, the accident occurrence time data may be the time point when the vehicle accident occurred. The accident geographic coordinates may be the longitude and latitude information of the vehicle accident site. The vehicle driving speed may be the numerical change in the speed of the target accident vehicle over a period of time before the vehicle accident occurred. The vehicle driving modes include pure electric, economic mode (the optimal amount of fuel is calculated based on intelligent comprehensive analysis of automatic transmission gear, engine speed, transmission oil temperature, etc. for engine use, thereby reducing fuel consumption) and standard mode (each performance of the car is stable and balanced, and the power performance and economic performance are kept in a balanced state). Traffic conditions may be information such as traffic conditions at the scene of the vehicle accident.
[0119] The present application obtains the accident occurrence time data by extracting the time data in the accident scene format data, obtains the accident geographic coordinates by extracting the geographic coordinates data in the accident scene format data, obtains the vehicle driving speed by extracting the vehicle status information in the accident scene format data, obtains the vehicle driving mode, and obtains the traffic road condition information in the accident scene format data to obtain the traffic condition.
[0120] It should be noted that the data contained in the accident scene format data may include data such as the weather conditions at the accident scene, the time of accident reporting, etc. in addition to the above-mentioned accident time data, accident geographic coordinates, vehicle driving speed, vehicle driving mode and traffic conditions.
[0121] In step S506 of some embodiments, the accident occurrence time data, accident geographic coordinates, vehicle driving speed, vehicle driving mode and traffic conditions are merged to obtain accident scene characteristic data.
[0122] In steps S501 to S506 shown in this embodiment, by extracting features from the accident scene format data, the accident time data, accident geographic coordinates, vehicle driving speed, vehicle driving mode and traffic conditions are obtained, which can improve the comprehensiveness of auto insurance anti-fraud detection and thus improve the accuracy of auto insurance anti-fraud detection.
[0123] In step S403 of some embodiments, after the accident scene characteristic data is obtained, the accident scene characteristic data is input into the accident data detection sub-model, and the gradient boosting decision tree algorithm contained in the accident data detection sub-model is used to calculate the probability that the accident scene characteristic data is accurate, and then an accident data detection result indicating whether the accident scene data is normal or abnormal is generated based on the probability that the accident scene characteristic data is accurate.
[0124] In steps S401 to S403 shown in the present embodiment, by performing steps such as format conversion and feature extraction on the accident scene data, the accident scene feature data required by the accident data detection sub-model for motor vehicle insurance anti-fraud detection is obtained, thereby reducing the impact of non-essential data in the accident scene data on the motor vehicle insurance anti-fraud detection results and improving the accuracy of motor vehicle insurance anti-fraud detection.
[0125] In step S302 of some embodiments, the target auto insurance data is input into an auto insurance data detection sub-model including a factor decomposition machine and a deep neural network, and an auto insurance anti-fraud detection is performed on the target auto insurance data to obtain an auto insurance data detection result.
[0126] It should be noted that the auto insurance data detection sub-model in this application adopts the DeepFM sorting model, which includes a factor decomposition machine and a deep neural network. The factor decomposition machine is responsible for learning low-order feature interactions, and the deep neural network is responsible for capturing high-order feature interactions.
[0127] For details, see Figure 6 In some embodiments, step S302 may include but is not limited to steps S601 to S604:
[0128] Step S601, based on the vehicle insurance data detection sub-model, feature extraction is performed on the vehicle insurance data to obtain vehicle insurance feature data;
[0129] Step S602, based on the factor decomposition machine, low-order feature interaction is performed on the auto insurance feature data to obtain low-order feature interaction data;
[0130] Step S603, based on the deep neural network, high-order feature interaction is performed on the auto insurance feature data to obtain high-order feature interaction data;
[0131] Step S604, determining the vehicle insurance data detection result based on the low-order feature interaction data and the high-order feature interaction data.
[0132] In step S601 of some embodiments, the feature extraction module in the vehicle insurance data detection sub-model is used to extract feature data in the target vehicle insurance data, and obtain vehicle insurance feature data that can be used for vehicle insurance data detection sub-model to perform vehicle insurance anti-fraud detection.
[0133] In one embodiment, the automobile insurance characteristic data may be automobile insurance policy amount data such as the total policy compensation amount, the vehicle damage compensation amount, etc.
[0134] In step S602 of some embodiments, a factor decomposition machine is used to perform simple linear combination or quadratic combination of features on the above-mentioned auto insurance feature data to obtain low-order feature interaction data.
[0135] In step S603 of some embodiments, a deep neural network is used to perform complex interactions between multiple features of the above-mentioned vehicle insurance feature data, for example, multi-layer nonlinear transformation, etc., to obtain high-order feature interaction data.
[0136] In step S604 of some embodiments, after obtaining the low-order feature interaction data and the high-order feature interaction data, the low-order feature interaction data and the high-order feature interaction data are combined to obtain a vehicle insurance data detection result indicating whether the target vehicle insurance data is normal or abnormal.
[0137] In steps S601 to S604 shown in the present embodiment, low-order feature interaction and high-order feature interaction are performed on the vehicle insurance feature data in the target vehicle insurance data to obtain low-order feature interaction data and high-order feature interaction data, and then the low-order feature interaction data is combined with the high-order feature interaction data to determine the vehicle insurance data detection result of the target vehicle insurance data, thereby improving the accuracy of the vehicle insurance data detection sub-model and thus improving the accuracy of the vehicle insurance data detection result.
[0138] In step S303 of some embodiments, after obtaining the auto insurance data detection result output by the auto insurance data detection sub-model, in order to further clarify whether the auto insurance report request sent by the client involves auto insurance fraud, it is also necessary to verify the auto insurance report request, and determine whether the accident scene data is normal or abnormal based on the accident data verification result obtained by verifying the auto insurance report request, which can improve the accuracy of auto insurance anti-fraud detection.
[0139] For details, see Figure 7 In some embodiments, step S303 may include but is not limited to steps S701 to S703:
[0140] Step S701, if the accident data detection result indicates that the accident scene data is normal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is normal, then generate investigation decision data, and determine the accident data verification result based on the investigation decision data;
[0141] Step S702, if the accident data detection result indicates that the accident scene data is abnormal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is abnormal, then generating an accident data verification result indicating that the accident scene data is abnormal;
[0142] Step S703: If the accident data detection result indicates that the accident scene data is abnormal, or the vehicle insurance data detection result indicates that the target vehicle insurance data is abnormal, then investigation decision data is generated, and based on the investigation decision data, the accident data verification result is determined.
[0143] In step S701 of some embodiments, the investigation decision data may be data obtained by the vehicle insurance compensation inspector during an investigation at the accident site, used to indicate whether the accident site data is true or false.
[0144] In the embodiment of the present application, when the accident data detection result indicates that the accident scene data is normal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is normal, the vehicle insurance compensation verifier can decide on his own whether it is necessary to dispatch staff to conduct on-site inspections, and generate inspection decision data based on the on-site inspection results of the staff. Based on the inspection decision data, it is determined whether the accident scene data is normal or abnormal, and the accident data verification result is obtained to ensure the authenticity and accuracy of the accident scene data.
[0145] In step S702 of some embodiments, when the accident data detection result indicates that the accident scene data is abnormal, and the auto insurance data detection result indicates that the target auto insurance data is abnormal, the embodiment of the present application can clearly state that the accident scene data must be abnormal. At this time, an accident data verification result indicating that the accident scene data is abnormal can be generated, reducing the steps of manual investigation, thereby improving the efficiency of auto insurance anti-fraud.
[0146] In step S703 of some embodiments, in the embodiments of the present application, when only the accident data detection result indicates that the accident scene data is abnormal, or only the motor vehicle insurance data detection result indicates that the target motor vehicle insurance data is abnormal, in order to ensure the accuracy of motor vehicle insurance anti-fraud, the auditor can dispatch staff to conduct on-site inspections, and generate inspection decision data based on the on-site inspection results of the staff, and then determine the accident data verification result based on the inspection decision data.
[0147] In steps S701 to S703 shown in this embodiment, by combining the motor vehicle insurance data detection results and the accident data detection results to verify whether the accident scene data is normal or abnormal, the robustness of the motor vehicle insurance anti-fraud model can be improved, thereby improving the accuracy of motor vehicle insurance anti-fraud detection, and reducing the time for manual accident scene inspection, thereby improving the efficiency of motor vehicle insurance anti-fraud detection.
[0148] In steps S301 to S303 shown in the present embodiment, an accident data detection sub-model is used to perform anti-fraud detection on the accident scene data to obtain an accident data detection result indicating whether the accident scene data is normal or abnormal. Then, based on the car insurance data detection sub-model, a car insurance fraud detection is performed on the target car insurance data to obtain a car insurance data detection result indicating whether the target car insurance data is normal or abnormal. This can achieve double verification of car insurance anti-fraud detection, thereby improving the accuracy of car insurance anti-fraud detection. Furthermore, based on the car insurance data detection result, the accident data detection result is verified to obtain an accident data verification result, which can further improve the accuracy of car insurance anti-fraud detection.
[0149] This application obtains a motor vehicle insurance anti-fraud model by training a preset machine learning model with vehicle accident data samples and motor vehicle insurance policy data samples. When a motor vehicle insurance report request carrying accident scene data is received from a client, the above-mentioned motor vehicle insurance anti-fraud model is used to perform anti-fraud detection on the accident scene data and the target motor vehicle insurance data of the target accident vehicle, which can reduce the need for manual investigation, reduce the consumption of human resources, and improve the efficiency of motor vehicle insurance anti-fraud detection.
[0150] See also Figure 8 The embodiment of the present application also provides a vehicle insurance anti-fraud device, which can implement the above vehicle insurance anti-fraud method, and the device includes:
[0151] The training data acquisition module 801 is used to acquire vehicle accident data samples and vehicle insurance policy data samples;
[0152] The machine model training module 802 is used to train a preset machine learning model based on vehicle accident data samples and auto insurance policy data samples to obtain an auto insurance anti-fraud model;
[0153] The accident vehicle confirmation module 803 is used to respond to the vehicle insurance report request with accident scene data sent by the client, and determine the target accident vehicle based on the accident scene data;
[0154] The insurance data acquisition module 804 is used to acquire the insurance data of the target accident vehicle from a preset insurance database to obtain the target insurance data;
[0155] The auto insurance anti-fraud detection module 805 is used to perform auto insurance anti-fraud detection on accident scene data and insurance data based on the auto insurance anti-fraud model.
[0156] The specific implementation of the vehicle insurance anti-fraud device is basically the same as the specific implementation of the above-mentioned vehicle insurance anti-fraud method, and will not be repeated here.
[0157] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned auto insurance anti-fraud method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.
[0158] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:
[0159] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;
[0160] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the auto insurance anti-fraud method of the embodiment of this application;
[0161] Input / output interface 903, used to implement information input and output;
[0162] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);
[0163] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);
[0164] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .
[0165] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the above-mentioned auto insurance anti-fraud method is implemented.
[0166] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0167] The motor vehicle insurance anti-fraud method, motor vehicle insurance anti-fraud device, electronic device and storage medium provided in the embodiments of the present application obtain a motor vehicle insurance anti-fraud model by training a preset machine learning model using vehicle accident data samples and motor vehicle insurance policy data samples, and upon receiving a motor vehicle insurance report request carrying accident scene data sent by a client, use the above-mentioned motor vehicle insurance anti-fraud model to perform anti-fraud detection on the accident scene data and the target motor vehicle insurance data of the target accident vehicle, thereby reducing the need for manual investigation, reducing the consumption of human resources, and improving the efficiency of motor vehicle insurance anti-fraud detection.
[0168] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.
[0169] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.
[0170] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0171] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.
[0172] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0173] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.
[0174] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0175] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0176] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0177] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.
[0178] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.
Claims
1. A method for anti-fraud of automobile insurance, characterized in that: Applied to the server, the method includes: Obtain vehicle accident data samples and vehicle insurance policy data samples; Based on the vehicle accident data sample and the auto insurance policy data sample, a preset machine learning model is trained to obtain an auto insurance anti-fraud model; In response to a vehicle insurance report request sent by a client and carrying accident scene data, a target accident vehicle is determined based on the accident scene data; Obtaining the vehicle insurance data of the target accident vehicle from a preset vehicle insurance database to obtain target vehicle insurance data; Based on the auto insurance anti-fraud model, auto insurance anti-fraud detection is performed on the accident scene data and the target auto insurance data.
2. The method according to claim 1, characterized in that: The auto insurance anti-fraud model includes an accident data detection sub-model and an auto insurance data detection sub-model. The auto insurance anti-fraud detection is performed on the accident scene data and the target auto insurance data based on the auto insurance anti-fraud model, including: Based on the accident data detection sub-model, the accident scene data is subjected to a motor vehicle insurance anti-fraud detection to obtain an accident data detection result, wherein the accident data detection result is used to indicate whether the accident scene data is normal or abnormal; Based on the vehicle insurance data detection sub-model, performing vehicle insurance anti-fraud detection on the target vehicle insurance data to obtain a vehicle insurance data detection result, wherein the vehicle insurance data detection result is used to indicate whether the target vehicle insurance data is normal or abnormal; Based on the vehicle insurance data detection result, the accident data detection result is verified to obtain an accident data verification result, wherein the accident data detection result is used to characterize whether the accident scene data is normal or abnormal.
3. The method according to claim 2, characterized in that The method of performing auto insurance anti-fraud detection on the accident scene data based on the accident data detection sub-model to obtain the accident data detection result includes: Converting the accident scene data into a format to obtain accident scene format data; Extracting features from the accident scene format data to obtain accident scene feature data; Based on the accident data detection sub-model and the gradient boosting decision tree algorithm, the accuracy of the accident scene feature data is estimated to obtain the accident data detection result.
4. The method according to claim 3, characterized in that The extracting features of the accident scene format data to obtain accident scene feature data includes: Performing time extraction on the accident scene format data to obtain accident occurrence time data; Extracting geographic information from the accident scene format data to obtain the accident geographic coordinates; Extracting the vehicle speed from the accident scene format data to obtain the vehicle driving speed; Extracting vehicle information from the accident scene format data to obtain a vehicle driving mode; Extracting traffic road condition information from the accident scene format data to obtain traffic conditions; The accident occurrence time data, the accident geographic coordinates, the vehicle driving speed, the vehicle driving mode and the traffic conditions are combined to obtain the accident scene characteristic data.
5. The method according to claim 2, characterized in that: The vehicle insurance data detection sub-model includes a factor decomposition machine and a deep neural network. Based on the vehicle insurance data detection sub-model, the target vehicle insurance data is subjected to vehicle insurance anti-fraud detection to obtain a vehicle insurance data detection result, including: Based on the vehicle insurance data detection sub-model, feature extraction is performed on the target vehicle insurance data to obtain vehicle insurance feature data; Based on the factor decomposition machine, performing low-order feature interaction on the auto insurance feature data to obtain low-order feature interaction data; Based on the deep neural network, high-order feature interaction is performed on the vehicle insurance feature data to obtain high-order feature interaction data; The vehicle insurance data detection result is determined based on the low-order feature interaction data and the high-order feature interaction data.
6. The method according to claim 2, characterized in that The step of verifying the accident data detection result based on the vehicle insurance data detection result to obtain the accident data verification result includes: If the accident data detection result indicates that the accident scene data is normal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is normal, then generating investigation decision data, and determining the accident data verification result based on the investigation decision data; If the accident data detection result indicates that the accident scene data is abnormal, and the vehicle insurance data detection result indicates that the target vehicle insurance data is abnormal, generating the accident data verification result indicating that the accident scene data is abnormal; If the accident data detection result indicates that the accident scene data is abnormal, or the vehicle insurance data detection result indicates that the target vehicle insurance data is abnormal, the investigation decision data is generated, and the accident data verification result is determined based on the investigation decision data.
7. The method according to any one of claims 1 to 6, characterized in that: The vehicle insurance database includes a vehicle information field and a vehicle insurance data field. The vehicle insurance data of the target accident vehicle is obtained from the preset vehicle insurance database to obtain the target vehicle insurance data, including: Acquiring vehicle information of the target accident vehicle; Based on the vehicle information, filtering out a target vehicle field from the vehicle information field; The vehicle insurance data field of the target vehicle field in the vehicle insurance database is queried to obtain the target vehicle insurance data.
8. A car insurance anti-fraud device, characterized in that: The device comprises: A training data acquisition module is used to obtain vehicle accident data samples and vehicle insurance policy data samples; A machine model training module, used to train a preset machine learning model based on the vehicle accident data sample and the motor insurance policy data sample to obtain a motor insurance anti-fraud model; An accident vehicle confirmation module, used to respond to a car insurance report request with accident scene data sent by a client, and determine a target accident vehicle based on the accident scene data; An insurance data acquisition module is used to acquire the insurance data of the target accident vehicle from a preset insurance database to obtain target insurance data; The auto insurance anti-fraud detection module is used to perform auto insurance anti-fraud detection on the accident scene data and the insurance data based on the auto insurance anti-fraud model.
9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the auto insurance anti-fraud method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the auto insurance anti-fraud method described in any one of claims 1 to 7 is implemented.
