Drunk driving fraud behavior identification method, device and equipment based on artificial intelligence

Through an artificial intelligence-based method, combining convolutional neural network and recurrent neural network model to process user tags and behavioral data, the problem of inefficient identification of drunk driving fraud in the existing technology is solved, and fast and accurate identification results are achieved.

CN119067684APending Publication Date: 2024-12-03CHINA PING AN PROPERTY INSURANCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411171994.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-12-03

Smart Images

  • Figure CN119067684A_ABST
    Figure CN119067684A_ABST
Patent Text Reader

Abstract

The invention belongs to the field of artificial intelligence and the field of financial science and technology, and relates to an artificial intelligence-based drunk driving fraud behavior identification method, which comprises the following steps of: obtaining user tag data and user behavior data of a user; preprocessing the user tag data and the user behavior data to obtain tag data and behavior data; performing feature extraction on the label data and the behavior data to obtain a first feature; integrating the first features to obtain second features; calling a first drunk driving fraud identification model and a second drunk driving fraud identification model to respectively identify the second feature to obtain a corresponding first prediction result and a corresponding second prediction result; and fusing the first prediction result and the second prediction result to generate a drunk driving fraud identification result of the user. The invention further provides a drunk driving fraud behavior recognition device based on artificial intelligence, computer equipment and a storage medium. In addition, the drunk driving fraud identification result can be stored in the block chain. According to the invention, the identification efficiency and the identification accuracy of drunk driving fraud behaviors are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence development technology and financial technology, and in particular to an artificial intelligence-based drunk driving fraud behavior identification method, device, computer equipment and storage medium. Background Art

[0002] With the rapid development of the transportation industry and the sharp increase in the number of motor vehicles, drunk driving has become a serious threat to road traffic safety worldwide. Drunk driving not only seriously violates traffic regulations, but also poses a great threat to the driver's own life safety and the life and property safety of other road users.

[0003] However, despite the increasingly stringent legal regulations applicable to combating and punishing drunk driving, some lawless elements still resort to various fraudulent means to cover up or falsify facts of drunk driving in order to evade legal responsibility or seek improper benefits. These fraudulent acts include but are not limited to: deliberately drinking alcohol to confuse the public after a traffic accident, falsifying evidence of drunk driving to defraud insurance compensation, using stand-ins or falsifying test results to evade investigation by traffic police, etc. These acts not only cause huge economic losses to insurance companies, but also seriously undermine social fairness and justice.

[0004] At present, insurance companies mainly rely on manual review and on-site investigation to identify drunk driving fraud. This identification method requires manual on-site investigation to obtain evidence, and relies on the auditor's professional knowledge and experience to generate corresponding drunk driving fraud identification results based on the evidence. Due to factors such as limited human resources and cumbersome review processes, it is easy to lead to low identification efficiency, time-consuming, and prone to misjudgment or omissions due to human factors. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose an artificial intelligence-based drunk driving fraud identification method, device, computer equipment and storage medium to solve the technical problems of the existing drunk driving fraud identification method adopted by insurance companies, which has low recognition efficiency, is time-consuming, and is prone to misjudgment or omission due to human factors.

[0006] In order to solve the above technical problems, the present application embodiment provides a method for identifying drunk driving fraud based on artificial intelligence, which adopts the following technical solutions:

[0007] Obtain user tag data and user behavior data of users;

[0008] Preprocessing the user tag data to obtain corresponding tag data, and preprocessing the user behavior data to obtain corresponding behavior data;

[0009] Extracting features from the label data and the behavior data to obtain corresponding first features; wherein the number of the first features includes a plurality;

[0010] Performing feature integration processing on all the first features to obtain a second feature;

[0011] Calling a pre-trained first drunk driving fraud identification model to perform identification processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud identification model is a model generated by training a convolutional neural network based on pre-collected sample data;

[0012] Calling a pre-trained second drunk driving fraud identification model to perform identification processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud identification model is a model generated by training a recurrent neural network based on the sample data;

[0013] The first prediction result and the second prediction result are fused based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user.

[0014] Furthermore, the step of fusing the first prediction result and the second prediction result based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user specifically includes:

[0015] Acquire a first recognition threshold corresponding to the first drunk driving fraud recognition model, and acquire a second recognition threshold corresponding to the second drunk driving fraud recognition model;

[0016] Performing data analysis on the first recognition threshold and the first prediction result to generate a first voting result corresponding to the first drunk driving fraud recognition model;

[0017] Performing data analysis on the second recognition threshold and the second prediction result to generate a second voting result corresponding to the second drunk driving fraud recognition model;

[0018] Determining whether the first voting result is the same as the second voting result;

[0019] If the first voting result is the same as the second voting result, the predicted result is used as the drunk driving fraud identification result; wherein the predicted result is the first predicted result or the second predicted result.

[0020] Furthermore, after the step of determining whether the first voting result is the same as the second voting result, the method further includes:

[0021] If the first voting result is different from the second voting result, determining whether there is a priority between the first drunk driving fraud identification model and the second drunk driving fraud identification model;

[0022] If there is a priority, selecting a first designated model with a higher priority from the first drunk driving fraud identification model and the second drunk driving fraud identification model;

[0023] Obtaining a first specified prediction result of the first specified model, and using the first specified prediction result as the drunk driving fraud identification result;

[0024] If there is no priority, obtaining a first preset weight of the first drunk driving fraud identification model, and obtaining a second preset weight of the second drunk driving fraud identification model;

[0025] Filtering a designated weight with a larger value from the first preset weight and the second preset weight, and determining a second designated model corresponding to the designated weight;

[0026] Obtain a second specified prediction result of the second specified model, and use the second specified prediction result as the drunk driving fraud identification result.

[0027] Furthermore, the step of preprocessing the user behavior data to obtain corresponding behavior data specifically includes:

[0028] Performing time series processing on the user behavior data to obtain corresponding first data;

[0029] Performing geographic location conversion processing on the user behavior data to obtain corresponding second data;

[0030] Performing behavior pattern extraction processing on the user behavior data to obtain corresponding third data;

[0031] Performing text extraction processing on the user behavior data to obtain corresponding fourth data;

[0032] The behavior data is generated based on the first data, the second data, the third data, and the fourth data.

[0033] Furthermore, before the step of calling the pre-trained first drunk driving fraud recognition model to perform recognition processing on the second feature to obtain the corresponding first prediction result, the method further includes:

[0034] Obtaining pre-collected initial sample data;

[0035] Performing sample enhancement processing on the initial sample data to obtain corresponding sample data;

[0036] Dividing the sample data into a training set and a test set according to a preset division ratio;

[0037] Calling a preset convolutional neural network, and training the convolutional neural network based on the training set to obtain a corresponding third specified model;

[0038] Testing the third specified model based on the test set;

[0039] If the third designated model passes the test, the third designated model is used as the first drunk driving fraud identification model.

[0040] Furthermore, the step of performing sample enhancement processing on the initial sample data to obtain corresponding sample data specifically includes:

[0041] Performing sample generation processing on the initial sample data based on a preset target oversampling method to obtain corresponding first sample data;

[0042] Call the pre-trained sample generation model;

[0043] Performing sample generation processing on the initial sample data based on the sample generation model to obtain corresponding second sample data;

[0044] The sample data is constructed based on the first sample data, the second sample data and the initial sample data.

[0045] Furthermore, the step of performing sample generation processing on the initial sample data based on a preset target oversampling method to obtain corresponding first sample data specifically includes:

[0046] Obtaining minority class samples from the initial sample data;

[0047] Performing interpolation processing on the minority class samples to obtain corresponding synthetic samples;

[0048] Integrate the synthesized sample and the initial sample data to obtain a processed integrated sample;

[0049] The integrated sample is used as the first sample data.

[0050] In order to solve the above technical problems, the embodiment of the present application also provides a drunk driving fraud behavior identification device based on artificial intelligence, which adopts the following technical solution:

[0051] The first acquisition module is used to acquire user tag data and user behavior data of the user;

[0052] A preprocessing module, used to preprocess the user label data to obtain corresponding label data, and to preprocess the user behavior data to obtain corresponding behavior data;

[0053] An extraction module, configured to extract features from the label data and the behavior data to obtain corresponding first features; wherein the number of the first features includes a plurality;

[0054] An integration module, used for performing feature integration processing on all the first features to obtain a second feature;

[0055] a first recognition module, configured to call a pre-trained first drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud recognition model is a model generated by training a convolutional neural network based on pre-collected sample data;

[0056] A second recognition module is used to call a pre-trained second drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud recognition model is a model generated by training a recurrent neural network based on the sample data;

[0057] A generation module is used to fuse the first prediction result and the second prediction result based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user.

[0058] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0059] The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the artificial intelligence-based drunk driving fraud behavior identification methods proposed in the embodiments of the present application are implemented.

[0060] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0061] The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of any one of the artificial intelligence-based drunk driving fraud behavior identification methods proposed in the embodiments of the present application.

[0062] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0063] In an embodiment of the present application, user label data and user behavior data of a user are obtained; the user label data is preprocessed to obtain corresponding label data, and the user behavior data is preprocessed to obtain corresponding behavior data; feature extraction is performed on the label data and the behavior data to obtain a corresponding first feature; wherein the number of the first features includes multiple; feature integration processing is performed on all the first features to obtain a second feature; a pre-trained first drunk driving fraud recognition model is called to perform recognition processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud recognition model is a model generated by training a convolutional neural network based on pre-collected sample data; a pre-trained second drunk driving fraud recognition model is called to perform recognition processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud recognition model is a model generated by training a recurrent neural network based on the sample data; the first prediction result and the second prediction result are fused based on a preset fusion strategy to generate a drunk driving fraud recognition result corresponding to the user. This application makes full use of the user's user label data and user behavior data, and combines the first drunk driving fraud identification model with the second drunk driving fraud identification model to identify and process the user label data and user behavior data, and then fuses the prediction results output by the first drunk driving fraud identification model and the second drunk driving fraud identification model based on the fusion strategy to generate a drunk driving fraud identification result corresponding to the user. Different from the existing identification method based on manual review and on-site investigation, this application can quickly and accurately identify drunk driving fraud, effectively improve the efficiency of drunk driving fraud identification, and improve the accuracy of the generated drunk driving fraud identification results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0065] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0066] Figure 2 A flowchart of an embodiment of a method for identifying drunk driving fraud based on artificial intelligence according to the present application;

[0067] Figure 3 It is a structural schematic diagram of an embodiment of an artificial intelligence-based drunk driving fraud behavior identification device according to the present application;

[0068] Figure 4It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0069] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0070] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0071] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0072] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0073] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0074] The terminal devices 101, 102, and 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, and the like.

[0075] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0076] It should be noted that the artificial intelligence-based drunk driving fraud behavior identification method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the artificial intelligence-based drunk driving fraud behavior identification device is generally set in the server / terminal device.

[0077] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0078] Continue to refer Figure 2 , showing a flow chart of an embodiment of a drunk driving fraud identification method based on artificial intelligence according to the present application. According to different needs, the order of the steps in the flow chart can be changed, and some steps can be omitted. The drunk driving fraud identification method based on artificial intelligence provided in the embodiment of the present application can be applied to any scenario that requires drunk driving fraud identification, and the drunk driving fraud identification method based on artificial intelligence can be applied to products in these scenarios, for example, drunk driving fraud identification in the financial insurance field. The drunk driving fraud identification method based on artificial intelligence includes the following steps:

[0079] Step S201, obtaining user tag data and user behavior data of the user.

[0080] In this embodiment, the method for identifying drunk driving fraud based on artificial intelligence is run on an electronic device (e.g. Figure 1The user tag data and user behavior data of the user can be obtained through a wired connection or a wireless connection. It should be noted that the above-mentioned wireless connection method may include but is not limited to 3G / 4G / 5G connection, Wi-Fi connection, Bluetooth connection, Wi MAX connection, Zigbee connection, UWB (Ultra Wideband) connection, and other wireless connection methods currently known or to be developed in the future. The above-mentioned users may be candidate users for insurance companies to conduct drunk driving fraud in handling auto insurance claims. The above-mentioned user tag data refers to the personal information provided by the user when registering or insuring with the insurance company, including but not limited to name, gender, age, occupation, address, contact information, vehicle information, etc. The above-mentioned user behavior data refers to the operation records or interaction information generated by the user on the insurance company platform, including but not limited to login time, login location, login device, browsing content, number of clicks, search keywords, insured products, insured amount, claim application, claim amount, claim evidence, etc.

[0081] Step S202: pre-processing the user tag data to obtain corresponding tag data, and pre-processing the user behavior data to obtain corresponding behavior data.

[0082] In this embodiment, the above-mentioned preprocessing of the user label data refers to cleaning, deduplication, missing value filling, and outlier processing operations on the user label data to improve the quality and consistency of the generated label data. Among them, data cleaning refers to the removal of duplicate, invalid or erroneous data to ensure the accuracy and completeness of the data. Missing value filling refers to the processing of missing values ​​by means of mean filling, interpolation, model prediction, etc. Outlier processing refers to the process of deleting outliers. In addition, the specific implementation process of the above-mentioned preprocessing of the user behavior data to obtain the corresponding behavior data will be further described in detail in the subsequent specific embodiments of this application, and will not be elaborated on here.

[0083] Step S203, extracting features from the label data and the behavior data to obtain corresponding first features; wherein the number of the first features includes multiple.

[0084] In this embodiment, the above-mentioned feature extraction refers to the operations of converting, encoding, normalizing, reducing dimension and the like on the label data and behavior data obtained after preprocessing, so as to extract effective features that can reflect user characteristics and behavior patterns. Specifically, effective features may at least include user portrait features, behavior model features, anomaly detection features and spatiotemporal features. Among them, user portrait features: construct user portrait features based on label data, such as age group, occupation category, residential area, etc. Behavior model features: extract behavior pattern features from behavior data, such as login frequency, browsing time, click preference, insurance habits, etc. Anomaly detection features: identify and extract abnormal features that do not match normal behavior patterns, such as frequent logins during non-working hours, multiple changes in insurance information in a short period of time, etc. Spatiotemporal features: extract spatiotemporal features such as user movement trajectories and frequently visited places, combined with geographic location and time information.

[0085] Step S204: performing feature integration processing on all the first features to obtain a second feature.

[0086] In this embodiment, the process of the feature integration processing includes: for the categorical features in the first feature (such as gender, occupation, geographic location classification, etc.), use an appropriate encoding method to convert them. For the numerical features in the first feature (such as age, amount, timestamp, etc.), normalization or standardization is performed to ensure that they have similar scales in the model. Subsequently, the encoded categorical features and the processed numerical features are concatenated into a long vector, and this long vector is used as the final feature vector, that is, the second feature. Among them, the above encoding method can use one-hot encoding, label encoding or embedding layer.

[0087] Step S205, calling a pre-trained first drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud recognition model is a model generated by training a convolutional neural network based on pre-collected sample data.

[0088] In this embodiment, the first prediction result is the judgment result of whether the user has drunk driving fraud behavior output by the first drunk driving fraud identification model. The first prediction result can be in the form of a probability value or a confidence score, which is used to indicate the reliability of the prediction result. Among them, the specific implementation process of the first drunk driving fraud identification model generated by training the convolutional neural network based on the pre-collected sample data will be further described in detail in the subsequent specific embodiments of this application, and will not be elaborated on here.

[0089] Step S206, calling a pre-trained second drunk driving fraud identification model to perform identification processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud identification model is a model generated by training a recurrent neural network based on the sample data.

[0090] In this embodiment, the second prediction result is a judgment result of whether the user has committed drunk driving fraud, output by the second drunk driving fraud identification model. The second prediction result can be in the form of a probability value or a confidence score, which is used to indicate the reliability of the prediction result. The training generation process of the second drunk driving fraud identification model can refer to the training generation process of the first drunk driving fraud identification model, and will not be described in detail here.

[0091] Step S207: The first prediction result and the second prediction result are fused based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user.

[0092] In this embodiment, the above-mentioned specific implementation process of fusing the first prediction result and the second prediction result based on the preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user will be further described in detail in the subsequent specific embodiments of this application and will not be elaborated here.

[0093] This application makes full use of the user's user label data and user behavior data, and combines the first drunk driving fraud identification model with the second drunk driving fraud identification model to identify and process the user label data and user behavior data, and then fuses the prediction results output by the first drunk driving fraud identification model and the second drunk driving fraud identification model based on the fusion strategy to generate a drunk driving fraud identification result corresponding to the user. Different from the existing identification method based on manual review and on-site investigation, this application can quickly and accurately identify drunk driving fraud, effectively improve the efficiency of drunk driving fraud identification, and improve the accuracy of the generated drunk driving fraud identification results.

[0094] In some optional implementations, step S207 includes the following steps:

[0095] A first recognition threshold corresponding to the first drunk driving fraud recognition model is obtained, and a second recognition threshold corresponding to the second drunk driving fraud recognition model is obtained.

[0096] In this embodiment, a recognition threshold is determined in advance for each of the first drunk driving fraud recognition model and the second drunk driving fraud recognition model. This recognition threshold is used to convert the continuous output of the model (such as probability value) into a discrete classification label (such as drunk driving / non-drunk driving). Among them, the selection of the recognition threshold is usually based on the performance of the model on the validation set. Specifically, the optimal recognition threshold can be determined by analyzing the ROC curve, the precision-recall curve, etc.

[0097] Data analysis is performed on the first recognition threshold and the first prediction result to generate a first voting result corresponding to the first drunk driving fraud recognition model.

[0098] In this embodiment, by comparing the first recognition threshold with the first prediction result, if the first prediction result is greater than the first recognition threshold, the corresponding first voting result is that there is drunk driving fraud. If the first prediction result is less than the first recognition threshold, the corresponding first voting result is that there is no drunk driving fraud.

[0099] Data analysis is performed on the second recognition threshold and the second prediction result to generate a second voting result corresponding to the second drunk driving fraud recognition model.

[0100] In this embodiment, by comparing the second recognition threshold with the second prediction result, if the second prediction result is greater than the second recognition threshold, the corresponding second voting result is that there is drunk driving fraud. If the second prediction result is less than the second recognition threshold, the corresponding first voting result is that there is no drunk driving fraud.

[0101] Determine whether the first voting result is the same as the second voting result.

[0102] In this embodiment, content similarity analysis is performed on the first voting result and the second voting result to determine whether the first voting result is the same as the second voting result.

[0103] If the first voting result is the same as the second voting result, the predicted result is used as the drunk driving fraud identification result; wherein the predicted result is the first predicted result or the second predicted result.

[0104] In this embodiment, if the first voting result of the first drunk driving fraud identification model is the same as the second voting result of the second drunk driving fraud identification model, the final decision is consistent with the voting result, that is, the first prediction result or the second prediction result is used as the above-mentioned drunk driving fraud identification result.

[0105] The present application obtains a first voting result by performing data analysis on a first recognition threshold and a first prediction result of a first drunk driving fraud recognition model, and obtains a second voting result by performing data analysis on a second recognition threshold and a second prediction result of a second drunk driving fraud recognition model, and then performs content analysis on the first voting result and the second voting result to generate a drunk driving fraud recognition result corresponding to the user, thereby achieving intelligent and accurate final decision making by combining the prediction results of the first drunk driving fraud recognition model and the second drunk driving fraud recognition model, effectively improving the recognition accuracy of drunk driving fraud recognition, and ensuring the accuracy of the generated drunk driving fraud recognition result.

[0106] In some optional implementations of this embodiment, after the step of determining whether the first voting result is the same as the second voting result, the electronic device may further perform the following steps:

[0107] If the first voting result is different from the second voting result, it is determined whether there is a priority between the first drunk driving fraud identification model and the second drunk driving fraud identification model.

[0108] In this embodiment, if the first voting result is different from the second voting result, a preset priority or a preset weight will be further selected to resolve the conflict. According to actual business needs, the first drunk driving fraud identification model and the second drunk driving fraud identification model can be prioritized or weighted in advance. Set priority: If a model is considered more reliable in a specific situation, its vote can be preset to have a higher priority. Set weight: The weight can be regarded as the number of votes, that is, more votes are allocated to more reliable models, that is, the larger the weight, the higher the reliability.

[0109] If there is a priority, a first designated model with a higher priority is selected from the first drunk driving fraud identification model and the second drunk driving fraud identification model.

[0110] In this embodiment, by obtaining the first priority of the first drunk driving fraud identification model and the second priority of the second drunk driving fraud identification model, and then performing a numerical comparison between the first priority and the second priority, the first designated model corresponding to the higher priority is screened out.

[0111] Obtain a first specified prediction result of the first specified model, and use the first specified prediction result as the drunk driving fraud identification result.

[0112] In this embodiment, by using the first designated prediction result of the first designated model with a higher priority as the drunk driving fraud identification result, the reliability and accuracy of the drunk driving fraud identification result are effectively improved.

[0113] If there is no priority, a first preset weight of the first drunk driving fraud identification model is obtained, and a second preset weight of the second drunk driving fraud identification model is obtained.

[0114] In this embodiment, when no priority is set for the first drunk driving fraud identification model and the second drunk driving fraud identification model, weights may be further set for the first drunk driving fraud identification model and the second drunk driving fraud identification model.

[0115] A designated weight with a larger value is selected from the first preset weight and the second preset weight, and a second designated model corresponding to the designated weight is determined.

[0116] In this embodiment, by comparing the first preset weight with the second preset weight, the second designated model corresponding to the larger weight value is screened out.

[0117] Obtain a second specified prediction result of the second specified model, and use the second specified prediction result as the drunk driving fraud identification result.

[0118] In this embodiment, by using the second designated prediction result of the second designated model with a greater weight as the drunk driving fraud identification result, the reliability and accuracy of the drunk driving fraud identification result are effectively improved.

[0119] When the present application detects that the first voting result is different from the second voting result, the preset priority or preset weight of the first drunk driving fraud identification model and the second drunk driving fraud identification model will be further obtained, and in the case of the preset priority, the first designated prediction result of the first designated model with a higher priority will be intelligently used as the drunk driving fraud identification result, effectively improving the reliability and accuracy of the drunk driving fraud identification result. Alternatively, in the case of the preset weight, the second designated prediction result of the second designated model with a greater weight will be intelligently used as the drunk driving fraud identification result, effectively improving the reliability and accuracy of the drunk driving fraud identification result.

[0120] In some optional implementations, the preprocessing of the user behavior data to obtain corresponding behavior data in step S202 includes the following steps:

[0121] The user behavior data is processed in a time series manner to obtain first data.

[0122] In this embodiment, the above-mentioned time serialization processing specifically includes: converting the time-related data contained in the user behavior data (such as login time, insurance time) into a unified time format, and extracting the corresponding time features (such as day of the week, hours).

[0123] The user behavior data is processed by geographic location conversion to obtain second data.

[0124] In this embodiment, the above-mentioned geographic location conversion processing specifically includes: converting the geographic location information such as login location, address, etc. contained in the user behavior data into longitude and latitude, or further converting it into geographic location classification, such as city, region.

[0125] The user behavior data is subjected to behavior pattern extraction processing to obtain third data.

[0126] In this embodiment, the above-mentioned behavior pattern extraction processing specifically includes: analyzing the behavior sequence in the user behavior data, and extracting frequent patterns, abnormal patterns or specific behavior sequences, such as multiple logins in a short period of time, quickly purchasing insurance after browsing specific content, etc.

[0127] The user behavior data is subjected to text extraction processing to obtain fourth data.

[0128] In this embodiment, the above text extraction process specifically includes: performing word segmentation, stop word removal, stem extraction and other processes on the search keywords and the text in the claim evidence.

[0129] The behavior data is generated based on the first data, the second data, the third data, and the fourth data.

[0130] In this embodiment, integrated data may be obtained by integrating the first data, the second data, the third data, and the fourth data, and the integrated data may be used as the behavior data.

[0131] The present application performs time serialization processing, geographic location conversion processing, behavior pattern extraction processing and text extraction processing on the user behavior data, thereby achieving fast and intelligent preprocessing of the user behavior data, effectively improving the data quality and consistency of the generated behavior data.

[0132] In some optional implementations, before step S205, the electronic device may further perform the following steps:

[0133] Get the pre-collected initial sample data.

[0134] In this embodiment, the above-mentioned initial sample data includes a certain number of data on drunk driving fraud and the corresponding content is a label indicating the existence of drunk driving fraud, and includes a certain number of data on non-drunk driving fraud and the corresponding content is a label indicating the absence of drunk driving fraud.

[0135] Perform sample enhancement processing on the initial sample data to obtain corresponding sample data.

[0136] In this embodiment, the above-mentioned sample enhancement processing is performed on the initial sample data to obtain the specific implementation process of the corresponding sample data. This application will further describe this in detail in the subsequent specific embodiments, and will not be elaborated on here. Among them, by performing sample enhancement processing on the initial sample data, the diversity of the sample data can be effectively enhanced. In addition, by using the sample data that has undergone sample enhancement processing to perform subsequent training processing of the first drunk driving fraud recognition model, the generalization ability and accuracy of the first drunk driving fraud recognition model can be effectively improved.

[0137] The sample data is divided into a training set and a test set according to a preset division ratio.

[0138] In this embodiment, there is no specific limitation on the numerical value selection of the above division ratio, which can be set according to the actual model training requirements, for example, it can be set to 7:3.

[0139] A preset convolutional neural network is called, and the convolutional neural network is trained based on the training set to obtain a corresponding third specified model.

[0140] In this embodiment, a convolutional neural network is pre-selected as the model architecture, and the weights and biases of the model are initialized, and the loss function of the model is defined as a cross entropy loss, and Adam is used as an optimizer. Specifically, the convolutional neural network is trained using the data in the training set. In each iteration, the feature vector is input into the convolutional neural network to calculate the prediction result, the loss value is calculated according to the loss function, and the model weight is updated using the optimizer until the convergence condition is met and the trained third specified model is obtained. Among them, the convergence condition may include reaching a preset number of iterations, and the value of the number of iterations can be set according to the actual training requirements. In addition, the corresponding data can be screened out from the initial sample data as a verification set, and the verification set can be used regularly during the training process to evaluate the model performance of the convolutional neural network, and the model parameters of the convolutional neural network (such as learning rate, batch size, number of iterations, etc.) are adjusted according to the verification results, and feature selection or feature engineering is performed to improve model performance.

[0141] The third specified model is tested based on the test set.

[0142] In this embodiment, the test set is input into the third specified model to evaluate the model performance of the third specified model using the test set, and the corresponding evaluation index is calculated. Then, it is determined whether the evaluation index is greater than the preset index threshold. If the evaluation index is greater than the index threshold, the third specified model is determined to have passed the test, otherwise the third specified model is determined to have failed the test. Among them, the above evaluation index may include any one of the indicators such as accuracy, precision, recall rate, F1 score, etc. The selection of the above index threshold can be set according to the actual business usage requirements.

[0143] If the third designated model passes the test, the third designated model is used as the first drunk driving fraud identification model.

[0144] In this embodiment, if the third designated model passes the test, it indicates that the third designated model is a model with expected performance, and the third designated model will be used as the first drunk driving fraud identification model. In addition, if the third designated model fails the test, the model parameters of the third designated model can be further adjusted according to the test results and the model can be retrained until a model with expected performance is obtained and used as the first drunk driving fraud identification model.

[0145] The present application obtains sample data by performing sample enhancement on the pre-collected initial sample data, and then divides the sample data into a training set and a test set for training and testing the convolutional neural network, thereby realizing the rapid and intelligent construction of a first drunk driving fraud recognition model that meets expectations, and effectively improving the generalization ability and accuracy of the first drunk driving fraud recognition model.

[0146] In some optional implementations of this embodiment, performing sample enhancement processing on the initial sample data to obtain corresponding sample data includes the following steps:

[0147] The initial sample data is processed for sample generation based on a preset target oversampling method to obtain corresponding first sample data.

[0148] In this embodiment, the above-mentioned specific implementation process of performing sample generation processing on the initial sample data based on the preset target oversampling method to obtain the corresponding first sample data will be further described in detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0149] Call the pre-trained sample generation model.

[0150] In this embodiment, the sample generation model is specifically a model based on a generative adversarial network. The sample generation model includes a generator and a discriminator. The generator is responsible for generating fake samples similar to real drunk driving fraud samples. The generator is usually a neural network, the input is random noise, and the output is the generated fake sample. The discriminator is responsible for distinguishing whether the input sample is real or a fake sample generated by the generator. The discriminator is also a neural network, the input is a sample, and the output is the probability that the sample is a real sample. The training process of the sample generation model includes: alternating training: in each round of training, the parameters of the generator are first fixed, and the discriminator is trained to distinguish between real samples and fake samples; then the parameters of the discriminator are fixed, and the generator is trained to generate more realistic fake samples, making it difficult for the discriminator to distinguish. Optimization goal: The goal of the generator is to generate fake samples that are as realistic as possible, so that the discriminator cannot distinguish; the goal of the discriminator is to distinguish between real samples and fake samples as accurately as possible. After completing the training of the sample generation model, a large number of fake samples are generated by using the sample generation model. These fake samples are similar to real samples but have diversity, which can be used to enhance the diversity of training data.

[0151] The initial sample data is processed for sample generation based on the sample generation model to obtain corresponding second sample data.

[0152] In this embodiment, the initial sample data is processed by using a sample generation model to generate a large number of false samples (i.e., second sample data). These false samples are similar to real samples (i.e., initial sample data) but have diversity, thereby effectively enhancing the diversity of sample data.

[0153] The sample data is constructed based on the first sample data, the second sample data and the initial sample data.

[0154] In this embodiment, integrated sample data may be obtained by integrating the first sample data, the second sample data and the initial sample data, and the integrated sample data may be used as final sample data.

[0155] The present application performs sample generation processing on the initial sample data using a target oversampling method to obtain corresponding first sample data; and calls a pre-trained sample generation model; performs sample generation processing on the initial sample data to obtain corresponding second sample data; and subsequently constructs the sample data based on the first sample data, the second sample data and the initial sample data, so as to effectively enhance the diversity of sample data by combining the use of the target oversampling method and the sample generation model, thereby facilitating the improvement of the generalization ability and accuracy of the drunk driving fraud recognition model generated by subsequent training.

[0156] In some optional implementations of this embodiment, performing sample generation processing on the initial sample data based on a preset target oversampling method to obtain corresponding first sample data includes the following steps:

[0157] Obtain minority class samples from the initial sample data.

[0158] In this embodiment, the minority class samples refer to the data of drunk driving fraud in the initial sample data.

[0159] Interpolation processing is performed on the minority class samples to obtain corresponding synthetic samples.

[0160] In this embodiment, the above interpolation processing specifically includes: randomly selecting a sample point from the minority class samples as a seed sample. Then in the feature space, find the k nearest neighbors of the seed sample. k is a preset parameter. Then generate a new sample: for each nearest neighbor, based on the difference between the seed sample and the nearest neighbor, a new sample is generated by calculating the difference between the two and multiplying it by a random number. This new sample is a synthetic point between the seed sample and the nearest neighbor. Repeat the above process: for each sample in the minority class samples, repeat the above steps to generate a specified number of new samples, thereby obtaining the above synthetic sample.

[0161] The synthetic sample and the initial sample data are integrated to obtain a processed integrated sample.

[0162] In this embodiment, the integrated sample includes the synthetic sample and the initial sample data.

[0163] The integrated sample is used as the first sample data.

[0164] The present application obtains minority class samples from the initial sample data; then interpolates the minority class samples to obtain corresponding synthetic samples; subsequently integrates the synthetic samples with the initial sample data to obtain processed integrated samples, and uses the integrated samples as the first sample data. The present application performs sample generation processing on the initial sample data by a target oversampling method, which can effectively enhance the diversity of the first sample data. In addition, the sample data is subsequently constructed based on the first sample data, the second sample data, and the initial sample data, which can effectively enhance the diversity of the sample data by combining the use of the target oversampling method and the sample generation model, thereby facilitating the improvement of the generalization ability and accuracy of the drunk driving fraud recognition model generated by subsequent training.

[0165] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0166] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned drunk driving fraud identification results, the above-mentioned drunk driving fraud identification results can also be stored in a blockchain node.

[0167] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0168] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0169] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0170] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a drunk driving fraud behavior identification device based on artificial intelligence, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0171] like Figure 3 As shown, the artificial intelligence-based drunk driving fraud behavior identification device 300 described in this embodiment includes: a first acquisition module 301, a preprocessing module 302, an extraction module 303, an integration module 304, a first identification module 305, a second identification module 306 and a generation module 307. Among them:

[0172] The first acquisition module 301 is used to acquire user tag data and user behavior data of the user;

[0173] A preprocessing module 302, configured to preprocess the user tag data to obtain corresponding tag data, and preprocess the user behavior data to obtain corresponding behavior data;

[0174] An extraction module 303 is used to extract features from the label data and the behavior data to obtain corresponding first features; wherein the number of the first features includes multiple;

[0175] An integration module 304 is used to perform feature integration processing on all the first features to obtain a second feature;

[0176] A first recognition module 305 is used to call a pre-trained first drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud recognition model is a model generated by training a convolutional neural network based on pre-collected sample data;

[0177] A second recognition module 306 is used to call a pre-trained second drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud recognition model is a model generated by training a recurrent neural network based on the sample data;

[0178] The generation module 307 is used to fuse the first prediction result and the second prediction result based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user.

[0179] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based drunk driving fraud behavior identification method in the aforementioned implementation method, and will not be repeated here.

[0180] In some optional implementations of this embodiment, the generating module 307 includes:

[0181] A first acquisition submodule, used to acquire a first recognition threshold corresponding to the first drunk driving fraud recognition model, and to acquire a second recognition threshold corresponding to the second drunk driving fraud recognition model;

[0182] A first generating submodule, configured to perform data analysis on the first recognition threshold and the first prediction result to generate a first voting result corresponding to the first drunk driving fraud recognition model;

[0183] A second generating submodule, configured to perform data analysis on the second recognition threshold and the second prediction result to generate a second voting result corresponding to the second drunk driving fraud recognition model;

[0184] A first judging submodule, used to judge whether the first voting result is the same as the second voting result;

[0185] The first determination submodule is used to use the predicted result as the drunk driving fraud identification result if the first voting result is the same as the second voting result; wherein the predicted result is the first predicted result or the second predicted result.

[0186] In some optional implementations of this embodiment, the generating module 307 further includes:

[0187] A second judgment submodule, configured to judge whether there is a priority between the first drunk driving fraud identification model and the second drunk driving fraud identification model if the first voting result is different from the second voting result;

[0188] a screening submodule, configured to screen out a first designated model with a higher priority from the first drunk driving fraud identification model and the second drunk driving fraud identification model if a priority exists;

[0189] A second acquisition submodule is used to acquire a first specified prediction result of the first specified model, and use the first specified prediction result as the drunk driving fraud identification result;

[0190] A third acquisition submodule, configured to acquire a first preset weight of the first drunk driving fraud identification model and acquire a second preset weight of the second drunk driving fraud identification model if no priority exists;

[0191] A second determination submodule, configured to select a designated weight with a larger value from the first preset weight and the second preset weight, and determine a second designated model corresponding to the designated weight;

[0192] The third determination submodule is used to obtain the second specified prediction result of the second specified model, and use the second specified prediction result as the drunk driving fraud identification result.

[0193] In some optional implementations of this embodiment, the preprocessing module 302 includes:

[0194] A first processing submodule, configured to perform time serialization processing on the user behavior data to obtain first data;

[0195] A second processing submodule, configured to perform geographic location conversion processing on the user behavior data to obtain second data;

[0196] A third processing submodule, configured to perform behavior pattern extraction processing on the user behavior data to obtain third data;

[0197] A fourth processing submodule, configured to perform text extraction processing on the user behavior data to obtain fourth data;

[0198] The third generating submodule is used to generate the behavior data based on the first data, the second data, the third data and the fourth data.

[0199] In some optional implementations of this embodiment, the artificial intelligence-based drunk driving fraud behavior identification device further includes:

[0200] A second acquisition module is used to acquire pre-collected initial sample data;

[0201] An enhancement module, used to perform sample enhancement processing on the initial sample data to obtain corresponding sample data;

[0202] A partitioning module, used to partition the sample data into a training set and a test set according to a preset partitioning ratio;

[0203] A training module, used to call a preset convolutional neural network, and train the convolutional neural network based on the training set to obtain a corresponding third specified model;

[0204] A testing module, used for testing the third specified model based on the test set;

[0205] A determination module is used to use the third designated model as the first drunk driving fraud identification model if the third designated model passes the test.

[0206] In some optional implementations of this embodiment, the enhancement module includes:

[0207] a fifth processing submodule, configured to perform sample generation processing on the initial sample data based on a preset target oversampling method to obtain corresponding first sample data;

[0208] The calling submodule is used to call the pre-trained sample generation model;

[0209] A sixth processing submodule, configured to perform sample generation processing on the initial sample data based on the sample generation model to obtain corresponding second sample data;

[0210] A construction submodule is used to construct the sample data based on the first sample data, the second sample data and the initial sample data.

[0211] In some optional implementations of this embodiment, the fifth processing submodule includes:

[0212] An acquisition unit, used for acquiring minority class samples from the initial sample data;

[0213] A processing unit, configured to perform interpolation processing on the minority class samples to obtain corresponding synthetic samples;

[0214] An integration unit, used for integrating the synthetic sample with the initial sample data to obtain an integrated sample after processing;

[0215] A determining unit is configured to use the integrated sample as the first sample data.

[0216] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0217] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital processors (DSPs), embedded devices, etc.

[0218] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0219] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Med ia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of a drunk driving fraud behavior identification method based on artificial intelligence, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0220] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the computer-readable instructions or process data stored in the memory 41, such as the computer-readable instructions for running the artificial intelligence-based drunk driving fraud identification method.

[0221] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0222] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the above-mentioned artificial intelligence-based drunk driving fraud behavior identification method.

[0223] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0224] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for identifying drunk driving fraud based on artificial intelligence, characterized in that: The steps include: Obtain user tag data and user behavior data of users; Preprocessing the user tag data to obtain corresponding tag data, and preprocessing the user behavior data to obtain corresponding behavior data; Extracting features from the label data and the behavior data to obtain corresponding first features; wherein the number of the first features includes a plurality; Performing feature integration processing on all the first features to obtain a second feature; Calling a pre-trained first drunk driving fraud identification model to perform identification processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud identification model is a model generated by training a convolutional neural network based on pre-collected sample data; Calling a pre-trained second drunk driving fraud identification model to perform identification processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud identification model is a model generated by training a recurrent neural network based on the sample data; The first prediction result and the second prediction result are fused based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user.

2. The artificial intelligence-based drunk driving fraud identification method according to claim 1 is characterized in that: The step of fusing the first prediction result and the second prediction result based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user specifically includes: Acquire a first recognition threshold corresponding to the first drunk driving fraud recognition model, and acquire a second recognition threshold corresponding to the second drunk driving fraud recognition model; Performing data analysis on the first recognition threshold and the first prediction result to generate a first voting result corresponding to the first drunk driving fraud recognition model; Performing data analysis on the second recognition threshold and the second prediction result to generate a second voting result corresponding to the second drunk driving fraud recognition model; Determining whether the first voting result is the same as the second voting result; If the first voting result is the same as the second voting result, the predicted result is used as the drunk driving fraud identification result; wherein the predicted result is the first predicted result or the second predicted result.

3. The artificial intelligence-based drunk driving fraud identification method according to claim 2 is characterized in that: After the step of determining whether the first voting result is the same as the second voting result, the method further includes: If the first voting result is different from the second voting result, determining whether there is a priority between the first drunk driving fraud identification model and the second drunk driving fraud identification model; If there is a priority, selecting a first designated model with a higher priority from the first drunk driving fraud identification model and the second drunk driving fraud identification model; Obtaining a first specified prediction result of the first specified model, and using the first specified prediction result as the drunk driving fraud identification result; If there is no priority, obtaining a first preset weight of the first drunk driving fraud identification model, and obtaining a second preset weight of the second drunk driving fraud identification model; Filtering a designated weight with a larger value from the first preset weight and the second preset weight, and determining a second designated model corresponding to the designated weight; Obtain a second specified prediction result of the second specified model, and use the second specified prediction result as the drunk driving fraud identification result.

4. The artificial intelligence-based drunk driving fraud identification method according to claim 1 is characterized in that: The step of preprocessing the user behavior data to obtain corresponding behavior data specifically includes: Performing time serialization processing on the user behavior data to obtain first data; Performing geographic location conversion processing on the user behavior data to obtain second data; Performing behavior pattern extraction processing on the user behavior data to obtain third data; Performing text extraction processing on the user behavior data to obtain fourth data; The behavior data is generated based on the first data, the second data, the third data, and the fourth data.

5. The artificial intelligence-based drunk driving fraud identification method according to claim 1 is characterized in that: Before the step of calling the pre-trained first drunk driving fraud identification model to identify the second feature and obtain the corresponding first prediction result, the step further includes: Obtaining pre-collected initial sample data; Performing sample enhancement processing on the initial sample data to obtain corresponding sample data; Dividing the sample data into a training set and a test set according to a preset division ratio; Calling a preset convolutional neural network, and training the convolutional neural network based on the training set to obtain a corresponding third specified model; Testing the third specified model based on the test set; If the third designated model passes the test, the third designated model is used as the first drunk driving fraud identification model.

6. The artificial intelligence-based drunk driving fraud identification method according to claim 5 is characterized in that: The step of performing sample enhancement processing on the initial sample data to obtain corresponding sample data specifically includes: Performing sample generation processing on the initial sample data based on a preset target oversampling method to obtain corresponding first sample data; Call the pre-trained sample generation model; Performing sample generation processing on the initial sample data based on the sample generation model to obtain corresponding second sample data; The sample data is constructed based on the first sample data, the second sample data and the initial sample data.

7. The artificial intelligence-based drunk driving fraud identification method according to claim 6 is characterized in that: The step of performing sample generation processing on the initial sample data based on a preset target oversampling method to obtain corresponding first sample data specifically includes: Obtaining minority class samples from the initial sample data; Performing interpolation processing on the minority class samples to obtain corresponding synthetic samples; Integrate the synthesized sample and the initial sample data to obtain a processed integrated sample; The integrated sample is used as the first sample data.

8. An artificial intelligence-based drunk driving fraud identification device, characterized in that: include: The first acquisition module is used to acquire user tag data and user behavior data of the user; A preprocessing module, used to preprocess the user label data to obtain corresponding label data, and to preprocess the user behavior data to obtain corresponding behavior data; An extraction module, configured to extract features from the label data and the behavior data to obtain corresponding first features; wherein the number of the first features includes a plurality; An integration module, used for performing feature integration processing on all the first features to obtain a second feature; a first recognition module, configured to call a pre-trained first drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding first prediction result; wherein the first drunk driving fraud recognition model is a model generated by training a convolutional neural network based on pre-collected sample data; A second recognition module is used to call a pre-trained second drunk driving fraud recognition model to perform recognition processing on the second feature to obtain a corresponding second prediction result; wherein the second drunk driving fraud recognition model is a model generated by training a recurrent neural network based on the sample data; A generation module is used to fuse the first prediction result and the second prediction result based on a preset fusion strategy to generate a drunk driving fraud identification result corresponding to the user.

9. A computer device, characterized in that: It includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the artificial intelligence-based drunk driving fraud behavior identification method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the artificial intelligence-based drunk driving fraud behavior identification method as described in any one of claims 1 to 7.