Automobile financial risk assessment method and system based on artificial intelligence

Abstract

The invention provides an automobile financial risk assessment method and system based on artificial intelligence, and relates to the technical field of artificial intelligence. The method comprises the following steps: firstly, outputting a financial risk description vector corresponding to target automobile financial risk data; secondly, merging the risk type, the enhanced description data corresponding to the risk type and the enhanced description data of the risk type with lower risk classification precision corresponding to the risk type to form risk type description data; then, outputting risk category description vectors corresponding to the risk category description data, and analyzing vector matching parameters between the financial risk description vector and the plurality of risk category description vectors; and finally, according to the risk type corresponding to the risk type description vector with the maximum value of the vector matching parameter, obtaining a target risk type. Based on the above content, the problems of relatively low reliability and efficiency of automobile financial risk assessment in the prior art can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an automobile financial risk assessment method and system based on artificial intelligence. Background Art

[0002] Auto finance generally includes financial services such as auto loans, among which there are generally various risk issues in financial services such as auto loans, such as various default issues, etc. Therefore, it is necessary to conduct risk assessment on the corresponding loan business, that is, to determine the corresponding risk type, etc. However, the inventors have found through research that in the prior art, the corresponding staff generally review and assess the relevant information collected from auto finance users to obtain the corresponding risk type, etc., which easily leads to the problem of relatively low reliability and efficiency of auto finance risk assessment. Summary of the invention

[0003] In view of this, the purpose of this application is to provide an automobile financial risk assessment method and system based on artificial intelligence to improve the problems of relatively low reliability and efficiency of automobile financial risk assessment in the prior art.

[0004] To achieve the above objectives, this application adopts the following technical solutions:

[0005] An automobile finance risk assessment method based on artificial intelligence, comprising:

[0006] Performing semantic mining on the target automobile financial risk data and outputting a financial risk description vector corresponding to the target automobile financial risk data, wherein the target automobile financial risk data is used to describe the automobile loan risk situation of the target automobile financial user, and the target automobile financial risk data at least includes basic user information and loan-related information of the target automobile financial user;

[0007] For each risk type among the multiple risk types with the highest risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with lower risk classification accuracy are merged to form risk type description data, wherein the enhanced description data is formed by learning the sample data, and the risk type with high risk classification accuracy and the corresponding risk type with low risk classification accuracy belong to the lower classification and the upper classification respectively;

[0008] Performing semantic mining on the risk type description data, outputting a risk type description vector corresponding to the risk type description data, and analyzing vector matching parameters respectively between the financial risk description vector and a plurality of risk type description vectors;

[0009] According to the risk type corresponding to the risk type description vector whose vector matching parameter has the maximum value among the multiple risk type description vectors, the target risk type corresponding to the target automobile financial risk data is obtained, wherein the target risk type corresponding to the target automobile financial risk data is used to reflect the risk type of the target automobile financial user for the corresponding automobile loan.

[0010] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, the step of performing semantic mining operation on the target automobile financial risk data and outputting the financial risk description vector corresponding to the target automobile financial risk data includes:

[0011] Perform word segmentation on the target automobile financial risk data to obtain multiple automobile financial risk words;

[0012] For each of the plurality of automobile financial risk words, performing word embedding processing on the automobile financial risk word to obtain a corresponding risk word embedding vector;

[0013] Based on the relationship between the multiple automobile financial risk words, a fusion operation is performed on the multiple risk word embedding vectors corresponding to the multiple automobile financial risk words to output a financial risk description vector corresponding to the target automobile financial risk data.

[0014] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, the step of fusing the multiple risk word embedding vectors corresponding to the multiple automobile financial risk words based on the relationship between the multiple automobile financial risk words and outputting the financial risk description vector corresponding to the target automobile financial risk data includes:

[0015] Performing mean superposition calculation on the risk word embedding vector corresponding to each word in the basic user information in the target automobile financial risk data, outputting the basic user information vector corresponding to the target automobile financial risk data, and performing mean superposition calculation on the risk word embedding vector corresponding to each word in the loan-related information in the target automobile financial risk data, outputting the loan-related information vector corresponding to the target automobile financial risk data;

[0016] Performing association semantic mining on the basic user information vector and the loan related information vector, and outputting a corresponding first association semantic mining vector;

[0017] Performing multi-level convolution operations on the basic user information vector and the loan-related information vector respectively, and outputting corresponding multiple basic user information convolution vectors and multiple loan-related information convolution vectors, wherein the convolution vector output by the previous level of convolution operation is used as the convolution vector input by the next level of convolution operation, and there is a one-to-one correspondence between the multiple basic user information convolution vectors and the multiple loan-related information convolution vectors;

[0018] For each basic user information convolution vector among the multiple basic user information convolution vectors, performing association semantic mining on the basic user information convolution vector and the corresponding loan-related information convolution vector, and outputting a corresponding second association semantic mining vector;

[0019] The first associated semantic mining vector and each of the second associated semantic mining vectors are concatenated to form a financial risk description vector corresponding to the target automobile financial risk data.

[0020] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, the step of obtaining the target risk type corresponding to the target automobile financial risk data according to the risk type corresponding to the risk type description vector having the maximum value of the vector matching parameter in the plurality of risk type description vectors comprises:

[0021] Determine the risk category corresponding to the risk category description vector having the maximum value of the vector matching parameter, and determine the risk category corresponding to the risk category with a lower risk classification accuracy;

[0022] The risk category corresponding to the risk category description vector having the maximum value of the vector matching parameter and the risk category corresponding to the risk category with a lower risk classification accuracy are marked to form a target risk category corresponding to the target automobile financial risk data.

[0023] In a preferred option of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, the risk types have a target number of risk classification accuracies;

[0024] Among them, for any two adjacent risk classification accuracies among the target number of risk classification accuracies: the a-th risk classification accuracy and the a+1-th risk classification accuracy, the a-th risk classification accuracy corresponds to a first preset number of risk types, the a+1-th risk classification accuracy corresponds to a second preset number of risk types, one or more risk types in the second preset number of risk types have lower risk classification accuracy and risk types with adjacent risk classification accuracies belong to the first preset number of risk types, the first preset number is not greater than the second preset number, and a+1 is less than or equal to the target number;

[0025] Furthermore, before executing the step of merging, for each of the multiple risk types with the highest risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with a lower risk classification accuracy corresponding to the risk type to form risk type description data, the automobile financial risk assessment method based on artificial intelligence further includes:

[0026] Performing semantic mining operations on the sample automobile financial risk data, and outputting a sample financial risk description vector corresponding to the sample automobile financial risk data;

[0027] For each risk classification accuracy of the target number of risk classification accuracies, determining vector matching parameters corresponding to the multiple risk types of the risk classification accuracy respectively according to the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy and the sample financial risk description vector;

[0028] Determining a first learning error corresponding to the risk classification accuracy according to vector matching parameters respectively corresponding to a plurality of risk types of the risk classification accuracy;

[0029] For the a+1th risk classification accuracy among the target number of risk classification accuracies, determine the second learning error corresponding to the ath risk classification accuracy according to the vector matching parameters corresponding to the multiple risk types of the a+1th risk classification accuracy and the vector matching parameters corresponding to the risk types of each risk type of the a+1th risk classification accuracy that has a lower risk classification accuracy and adjacent risk classification accuracy;

[0030] Obtaining a target learning error according to the first learning error and the second learning error;

[0031] According to the target learning error, the representation parameters of the reinforcement description data corresponding to multiple risk types of each risk classification accuracy among the target number of risk classification accuracies are optimized to form the reinforcement description data after learning, wherein the representation parameters of the initial reinforcement description data are randomly generated.

[0032] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, for each of the target number of risk classification accuracies, the step of determining the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy respectively according to the sample risk type description vectors and the sample financial risk description vectors corresponding to the multiple risk types of the risk classification accuracy respectively includes:

[0033] For each risk classification accuracy of the target number of risk classification accuracies, according to the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies, mining the sample risk category description vectors corresponding to the multiple risk categories of the risk classification accuracy;

[0034] Based on the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy, the sample financial risk description vectors and preset adjustment parameters, the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy are determined.

[0035] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, for each of the target number of risk classification accuracies, according to the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies, the step of mining the sample risk category description vectors corresponding to the multiple risk categories of the risk classification accuracy includes:

[0036] For each risk classification accuracy among the target number of risk classification accuracies, when the risk classification accuracy does not belong to the highest risk classification accuracy among the target number of risk classification accuracies, for each risk type among multiple risk types of the risk classification accuracy, a merging operation is performed on the risk type and the enhanced description data corresponding to the risk type to form corresponding sample risk type description data;

[0037] A semantic mining operation is performed on the sample risk type description data to output a sample risk type description vector corresponding to the sample risk type description data.

[0038] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, for each of the target number of risk classification accuracies, according to the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies, the step of mining the sample risk category description vectors corresponding to the multiple risk categories of the risk classification accuracy includes:

[0039] For each risk classification accuracy among the target number of risk classification accuracy, when the risk classification accuracy belongs to the highest risk classification accuracy among the target number of risk classification accuracy, for each risk type among multiple risk types of the risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with a lower risk classification accuracy of the risk type are merged to form corresponding sample risk type description data;

[0040] A semantic mining operation is performed on the sample risk type description data to output a sample risk type description vector corresponding to the sample risk type description data.

[0041] In a preferred embodiment of the present application, in the above-mentioned automobile financial risk assessment method based on artificial intelligence, the step of determining the second learning error corresponding to the a+1th risk classification accuracy among the target number of risk classification accuracies according to the vector matching parameters corresponding to the multiple risk types of the a+1th risk classification accuracy and the vector matching parameters corresponding to the risk types with lower risk classification accuracy and adjacent risk classification accuracy of each risk type of the a+1th risk classification accuracy, includes:

[0042] Constructing a comparative risk classification accuracy corresponding to the a-th risk classification accuracy, wherein the number of risk types of the comparative risk classification accuracy is the same as the number of risk types of the a+1-th risk classification accuracy, and there is a one-to-one correspondence between the risk types of the comparative risk classification accuracy and the risk types of the a+1-th risk classification accuracy;

[0043] For each risk type of the a+1th risk classification accuracy, obtain the vector matching parameters of the risk type corresponding to the comparative risk classification accuracy of the risk type according to the vector matching parameters of the risk types with lower risk classification accuracy and adjacent risk classification accuracy of the risk type;

[0044] The second learning error corresponding to the a-th risk classification accuracy is determined according to the vector matching parameters of each risk type of the a+1-th risk classification accuracy and the vector matching parameters of the risk type corresponding to the comparative risk classification accuracy.

[0045] On the basis of the above, the present application also provides an automobile financial risk assessment method system based on artificial intelligence, including:

[0046] Memory for storing computer programs;

[0047] A processor connected to the memory is used to execute the computer program stored in the memory to implement the above-mentioned automobile financial risk assessment method based on artificial intelligence.

[0048] The automobile financial risk assessment method and system based on artificial intelligence provided by the present application first outputs the financial risk description vector corresponding to the target automobile financial risk data; secondly, the risk type, the enhanced description data corresponding to the risk type and the enhanced description data of the risk type corresponding to the risk type with lower risk classification accuracy are merged to form risk type description data; then, the risk type description vector corresponding to the risk type description data is output, and the vector matching parameters respectively possessed by the financial risk description vector and the multiple risk type description vectors are analyzed; finally, the target risk type is obtained according to the risk type corresponding to the risk type description vector with the maximum value of the vector matching parameter. Based on the above content, since the risk type description data includes not only the corresponding risk type, but also the enhanced description data corresponding to the risk type, and also includes the enhanced description data of the risk type corresponding to the risk type with lower risk classification accuracy, the semantic information of the mined risk type description vector can be enriched, thereby providing a more reliable basis for subsequent analysis, so that the reliability of the determined risk type can also be higher, and therefore, the problem of relatively low reliability and efficiency of automobile financial risk assessment existing in the prior art can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings.

[0050] Figure 1 A structural block diagram of an artificial intelligence-based automobile financial risk assessment system provided in an embodiment of the present application.

[0051] Figure 2 A flowchart of an artificial intelligence-based automobile financial risk assessment method provided in an embodiment of the present application.

[0052] Figure 3 A schematic diagram of mining a financial risk description vector provided in an embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0055] like Figure 1 As shown, an embodiment of the present application provides an automobile financial risk assessment system based on artificial intelligence. Among them, the automobile financial risk assessment system based on artificial intelligence may include a memory and a processor. In detail, the memory and the processor are directly or indirectly electrically connected to realize data transmission or interaction. For example, the memory and the processor can be electrically connected through one or more communication buses or signal lines. The processor is used to execute an executable computer program stored in the memory to implement the automobile financial risk assessment method based on artificial intelligence provided in the embodiment of the present application.

[0056] Optionally, the memory can be, but is not limited to, random access memory (RAM), read only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable read-only memory (Electric Erasable Programmable Read-Only Memory, EEPROM), etc.

[0057] Furthermore, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0058] Understandably, Figure 1 The structure shown is for illustration only. The automobile financial risk assessment system based on artificial intelligence may also include Figure 1 More or fewer components as shown, or with Figure 1The different configurations shown, for example, may also include a communication unit for information exchange with other devices. In addition, the automobile financial risk assessment system based on artificial intelligence may be a server.

[0059] Combination Figure 2 The embodiment of the present application also provides an artificial intelligence-based automobile financial risk assessment method that can be applied to the above-mentioned artificial intelligence-based automobile financial risk assessment system. Among them, the method steps defined in the process related to the artificial intelligence-based automobile financial risk assessment method can be implemented by the artificial intelligence-based automobile financial risk assessment system (hereinafter referred to as the risk assessment system). Figure 2 The specific process shown is explained in detail.

[0060] Step S110 , performing semantic mining operation on the target automobile financial risk data, and outputting a financial risk description vector corresponding to the target automobile financial risk data.

[0061] In an embodiment of the present application, the risk assessment system may perform semantic mining operations on the target automobile financial risk data, and output a financial risk description vector corresponding to the target automobile financial risk data, wherein the financial risk description vector is used to characterize the semantic information in the target automobile financial risk data. The target automobile financial risk data is used to describe the automobile loan risk situation of the target automobile financial user, and the target automobile financial risk data at least includes the basic user information (such as gender, age, occupation, etc.) and loan-related information (such as credit score, income, daily expenses, family situation, existing assets, liabilities, etc.) of the target automobile financial user.

[0062] Step S120, for each risk type among the multiple risk types with the highest risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with lower risk classification accuracy are merged to form risk type description data.

[0063] In an embodiment of the present application, the risk assessment system may merge the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with lower risk classification accuracy (which may include enhanced description data corresponding to each risk type with lower risk classification accuracy) for each risk type among the multiple risk types with the highest risk classification accuracy, to form risk type description data. The enhanced description data is formed by learning the sample data, and the risk types with high risk classification accuracy and the corresponding risk types with low risk classification accuracy belong to lower-level classifications and higher-level classifications, respectively. For example, for an auto loan business, a risk type may have three risk classification accuracies, from low to high, namely the first risk classification accuracy, the second risk classification accuracy, and the third risk classification accuracy. Specific examples of the three risk classification accuracies are shown in the following table:

[0064]

[0065] Step S130, performing semantic mining operations on the risk type description data, outputting a risk type description vector corresponding to the risk type description data, and analyzing vector matching parameters respectively between the financial risk description vector and a plurality of risk type description vectors.

[0066] In an embodiment of the present application, after forming the risk type description data, the risk assessment system can perform semantic mining operations on the risk type description data, output the risk type description vector corresponding to the risk type description data, and analyze the vector matching parameters respectively possessed by the financial risk description vector and the plurality of risk type description vectors. Among them, the risk type description vector can be used to characterize the semantic information possessed by the risk type description data. Exemplarily, when the risk type description data is subjected to semantic mining operations, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with lower risk classification accuracy corresponding to the risk type can be respectively subjected to word embedding processing, and then, the mean of the word embedding vectors can be respectively calculated, and finally, each mean vector can be spliced ​​to obtain the risk type description vector corresponding to the risk type description data. In addition, in some embodiments, only the risk type can be subjected to word embedding processing, and the mean of the embedding vector corresponding to each word can be calculated to obtain the mean vector, and then, the mean vector and the characterization parameters corresponding to each enhanced description data (which can be in the form of a vector) can be directly spliced ​​to obtain the risk type description vector. In addition, when determining the vector matching parameters, the cosine similarity between the vectors may be calculated.

[0067] Step S140 , obtaining a target risk category corresponding to the target automobile financial risk data according to the risk category corresponding to the risk category description vector having the maximum vector matching parameter among the plurality of risk category description vectors.

[0068] In the embodiment of the present application, after forming the risk type description vector and the risk type description vector, the risk assessment system can obtain the target risk type corresponding to the target automobile financial risk data according to the risk type corresponding to the risk type description vector with the maximum value of the vector matching parameter in the plurality of risk type description vectors. The target risk type corresponding to the target automobile financial risk data is used to reflect the risk type of the target automobile financial user for the corresponding automobile loan, such as default - long-term overdue default - extremely high degree of overdue.

[0069] Based on the above content, since the risk type description data not only includes the corresponding risk types, but also includes the enhanced description data corresponding to the risk types, and also includes the enhanced description data of the risk types corresponding to the risk types with lower risk classification accuracy, the semantic information of the mined risk type description vector can be richer, thereby providing a more reliable basis for subsequent analysis, and the reliability of the determined risk types can also be higher. Therefore, the problems of relatively low reliability and efficiency of automobile financial risk assessment in the prior art can be improved.

[0070] It should be noted that for step S110 , the specific method of performing semantic mining operation on the target automobile financial risk data is not limited and can be selected according to actual needs.

[0071] For example, in an alternative implementation, in order to ensure that the mined financial risk description vector has better semantic representation ability, the above-mentioned step S110 may further include step S111, step S112 and step S113, and the specific content of each step is as follows.

[0072] Step S111 , performing word segmentation processing on the target automobile financial risk data to obtain a plurality of automobile financial risk words.

[0073] In an embodiment of the present application, the target automobile financial risk data may be segmented to obtain a plurality of automobile financial risk words, such as "age", "gender", "occupation", etc.

[0074] Step S112: for each automobile financial risk word among the plurality of automobile financial risk words, perform word embedding processing on the automobile financial risk word to obtain a corresponding risk word embedding vector.

[0075] In the embodiment of the present application, after obtaining a plurality of automobile financial risk words, a word embedding process can be performed on each of the plurality of automobile financial risk words to obtain a corresponding risk word embedding vector. For example, word embedding processes can be performed on "age", "gender" and "occupation" respectively to obtain a corresponding risk word embedding vector, which can be processed by a trained word embedding model.

[0076] Step S113, based on the relationship between the multiple automobile financial risk words, a fusion operation is performed on the multiple risk word embedding vectors corresponding to the multiple automobile financial risk words, and a financial risk description vector corresponding to the target automobile financial risk data is output.

[0077] In the embodiment of the present application, after the risk word embedding vector is obtained, a fusion operation can be performed on the multiple risk word embedding vectors corresponding to the multiple automobile financial risk words based on the relationship between the multiple automobile financial risk words, and the financial risk description vector corresponding to the target automobile financial risk data can be output. In this way, the financial risk description vector can fully and reliably represent the semantic information of each automobile financial risk word.

[0078] It is understandable that in the above step S113, the specific method of fusing the multiple risk word embedding vectors corresponding to the multiple automobile financial risk words is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to ensure that the obtained financial risk description vector has a high reliability, the above step S113 may include the following specific implementation content (which can be combined with Figure 3 shown):

[0079] First, the risk word embedding vector corresponding to each word in the basic user information in the target automobile financial risk data can be averaged and superimposed to output the basic user information vector corresponding to the target automobile financial risk data, and the risk word embedding vector corresponding to each word in the loan-related information in the target automobile financial risk data can be averaged and superimposed to output the loan-related information vector corresponding to the target automobile financial risk data;

[0080] Secondly, the basic user information vector and the loan-related information vector may be subjected to association semantic mining to output a corresponding first association semantic mining vector; illustratively, the basic user information vector and the transposed vector of the loan-related information vector may be multiplied, and then the multiplication result may be multiplied with the loan-related information vector, and finally, the multiplication result, the basic user information vector and the loan-related information vector may be subjected to mean or weighted sum calculation to obtain the corresponding first association semantic mining vector. If a weighted sum calculation is performed, the corresponding weight coefficient may be formed in the process of learning and training using the corresponding sample data;

[0081] Then, the basic user information vector and the loan-related information vector can be subjected to multi-level convolution operations respectively, and a corresponding plurality of basic user information convolution vectors and a plurality of loan-related information convolution vectors can be output, wherein the convolution vector output by the convolution operation of the previous level is used as the convolution vector input by the convolution operation of the next level, and there is a one-to-one correspondence between the plurality of basic user information convolution vectors and the plurality of loan-related information convolution vectors; in addition, if the dimensions of the vector before the convolution are different from those after the convolution, they can be processed by edge filling so that the vector dimensions of the plurality of basic user information convolution vectors and the plurality of loan-related information convolution vectors are the same, which can facilitate subsequent processing;

[0082] Furthermore, for each of the multiple basic user information convolution vectors, association semantic mining (as described above) may be performed on the basic user information convolution vector and the corresponding loan-related information convolution vector (the loan-related information convolution vector at the same level), and a corresponding second association semantic mining vector may be output;

[0083] Finally, the first associated semantic mining vector and each of the second associated semantic mining vectors may be concatenated to form a financial risk description vector corresponding to the target automobile financial risk data.

[0084] What needs to be explained about step S120 is that, when forming the risk type description data, not only the corresponding risk type will be used, but also the corresponding enhanced description data will be used. Therefore, it is necessary to determine the enhanced description data. Therefore, in order to make the enhanced description data able to reliably characterize the corresponding risk type, before executing step S120, the artificial intelligence-based automobile financial risk assessment method may also include the following steps S101, step S102, step S103, step S104, step S105 and step S106. Furthermore, the risk types have a target number of risk classification accuracies (such as the aforementioned 3), and for any two adjacent risk classification accuracies among the target number of risk classification accuracies: the ath risk classification accuracy and the a+1th risk classification accuracy (such as the second risk classification accuracy and the third risk classification accuracy), the ath risk classification accuracy corresponds to a first preset number of risk types (such as the aforementioned 3), the a+1th risk classification accuracy corresponds to a second preset number of risk types (such as the aforementioned 7), one or more risk types among the second preset number of risk types have lower risk classification accuracy and risk types with adjacent risk classification accuracies belong to the first preset number of risk types, the first preset number is not greater than the second preset number, and a+1 is less than or equal to the target number.

[0085] Step S101 , performing semantic mining operations on sample automobile financial risk data, and outputting sample financial risk description vectors corresponding to the sample automobile financial risk data.

[0086] In the embodiment of the present application, the sample automobile financial risk data may be subjected to semantic mining operations to output a sample financial risk description vector corresponding to the sample automobile financial risk data. The sample automobile financial risk data may refer to the relevant explanations of the target automobile financial risk data, and the process of performing the semantic mining operation may refer to the explanation of step S110.

[0087] Step S102, for each risk classification accuracy of the target number of risk classification accuracies, according to the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy and the sample financial risk description vector, determine the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy.

[0088] In an embodiment of the present application, after obtaining the sample financial risk description vector, for each of the target number of risk classification accuracies, the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy can be determined based on the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy and the sample financial risk description vector. The mining method of the sample risk type description vector can refer to the method of performing semantic mining operations on the risk type description data in the previous text.

[0089] Step S103: determining a first learning error corresponding to the risk classification accuracy according to vector matching parameters corresponding to a plurality of risk types of the risk classification accuracy.

[0090] In an embodiment of the present application, after obtaining the vector matching parameters, the first learning error corresponding to the risk classification accuracy can be determined based on the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy. Exemplarily, the vector matching parameters corresponding to the multiple risk types can constitute a vector matching parameter distribution matrix, and then a parameter matrix can be constructed based on the actual risk types of the risk type description data. In the parameter matrix, the parameters of the positions corresponding to the actual risk types are equal to 1, and the parameters of other positions are equal to 0. In this way, the mean square error between the parameter matrix and the vector matching parameter distribution matrix can be calculated as the first learning error.

[0091] Step S104, for the a+1th risk classification accuracy among the target number of risk classification accuracy, determine the second learning error corresponding to the a+1th risk classification accuracy based on the vector matching parameters corresponding to multiple risk types of the a+1th risk classification accuracy and the vector matching parameters corresponding to risk types with lower risk classification accuracy and adjacent risk classification accuracy of each risk type of the a+1th risk classification accuracy.

[0092] In the embodiment of the present application, after obtaining the vector matching parameters, the second learning error corresponding to the a+1th risk classification accuracy among the target number of risk classification accuracies can be determined based on the vector matching parameters corresponding to the multiple risk types of the a+1th risk classification accuracy and the vector matching parameters corresponding to the risk types of each risk type of the a+1th risk classification accuracy with a lower risk classification accuracy and adjacent risk classification accuracy (i.e., the risk type corresponding to the a+1th risk classification accuracy). In other words, the risk types corresponding to two adjacent risk classification accuracies can be compared and analyzed to determine the second learning error.

[0093] Step S105: obtaining a target learning error according to the first learning error and the second learning error.

[0094] In an embodiment of the present application, after the first learning error and the second learning error, a target learning error can be obtained according to the first learning error and the second learning error. Exemplarily, the first learning error and the second learning error can be summed or weighted summed to obtain a corresponding target learning error.

[0095] Step S106, based on the target learning error, optimizing the representation parameters of the enhanced description data corresponding to the multiple risk types of each risk classification accuracy in the target number of risk classification accuracies, to form enhanced description data after learning.

[0096] In an embodiment of the present application, after obtaining the target learning error, the characterization parameters of the reinforcement description data corresponding to the multiple risk types of each risk classification accuracy in the target number of risk classification accuracies can be optimized according to the target learning error to form the reinforcement description data after learning. Among them, the characterization parameters of the initial reinforcement description data are randomly generated. Exemplarily, the characterization parameters of the reinforcement description data can be updated in the direction of reducing the target learning error, so that the target learning error is reduced to a preset value, thereby obtaining the reinforcement description data after learning. In addition, the characterization parameters of the reinforcement description data can be the network parameters of the corresponding neural network, and the neural network can also include the aforementioned part corresponding to the semantic mining operation, and the network parameters of this part can also be updated and trained together.

[0097] It is understandable that, in the above step S102, the specific method of determining the vector matching parameters corresponding to the multiple risk types of risk classification accuracy is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to ensure that the determined vector matching parameters have a high reliability, the above step S102 can further include step S102a and step S102b, and the specific content of each step is as follows.

[0098] Step S102a, for each risk classification accuracy of the target number of risk classification accuracies, according to the upper and lower distribution relationship of the risk classification accuracy with respect to the target number of risk classification accuracies, mining out sample risk category description vectors corresponding to multiple risk categories of the risk classification accuracy.

[0099] In an embodiment of the present application, for each risk classification accuracy among the target number of risk classification accuracies, sample risk type description vectors corresponding to multiple risk types of the risk classification accuracy can be mined based on the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies.

[0100] Step S102b, determining vector matching parameters corresponding to the multiple risk types of the risk classification accuracy respectively according to the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy respectively, the sample financial risk description vectors and preset adjustment parameters.

[0101] In an embodiment of the present application, after obtaining the corresponding sample risk type description vector, the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy can be determined based on the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy, the sample financial risk description vector and the preset adjustment parameters (which can be configured according to actual needs). Exemplarily, the vector inner product between each sample risk type description vector and the sample financial risk description vector can be calculated respectively, and then the ratio between the vector inner product and the adjustment parameter is calculated, and then an exponential operation can be performed based on the ratio, and finally, the results of each exponential operation are normalized to obtain the vector matching parameters corresponding to each risk type.

[0102] It can be understood that, in the above step S102a, the specific method of mining the sample risk category description vectors corresponding to the multiple risk categories of the risk classification accuracy is not limited and can be selected according to actual needs.

[0103] For example, in an alternative implementation, in order to ensure that the semantic representation accuracy of the mined sample risk category description vector is relatively higher, the above step S102a may further include the following specific implementation contents:

[0104] First, for each risk classification accuracy among the target number of risk classification accuracy, when the risk classification accuracy does not belong to the highest risk classification accuracy among the target number of risk classification accuracy, for each risk type among multiple risk types of the risk classification accuracy, the risk type and the enhanced description data corresponding to the risk type may be merged to form corresponding sample risk type description data;

[0105] Secondly, a semantic mining operation is performed on the sample risk type description data to output a sample risk type description vector corresponding to the sample risk type description data, and reference may be made to the above description.

[0106] For example, in another alternative implementation, in order to ensure that the semantic representation accuracy of the mined sample risk type description vector is relatively higher, the above step S102a may further include the following specific implementation contents:

[0107] First, for each risk classification accuracy among the target number of risk classification accuracy, when the risk classification accuracy belongs to the highest risk classification accuracy among the target number of risk classification accuracy, for each risk type among multiple risk types of the risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with a lower risk classification accuracy of the risk type are merged to form corresponding sample risk type description data;

[0108] Secondly, a semantic mining operation is performed on the sample risk type description data to output a sample risk type description vector corresponding to the sample risk type description data, and reference may be made to the above description.

[0109] It can be understood that, in the above step S104, the specific method of determining the second learning error corresponding to the a-th risk classification accuracy is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to ensure that the determined second learning error can better represent the degree of optimization and updating in the learning process, the above step S104 may include:

[0110] First, a comparative risk classification accuracy corresponding to the a-th risk classification accuracy can be constructed, wherein the number of risk types of the comparative risk classification accuracy is the same as the number of risk types of the a+1-th risk classification accuracy, and there is a one-to-one correspondence between the risk types of the comparative risk classification accuracy and the risk types of the a+1-th risk classification accuracy; for example, the risk types of the a+1-th risk classification accuracy can be used as the risk types of the comparative risk classification accuracy;

[0111] Secondly, for each risk type of the a+1th risk classification accuracy, the vector matching parameters of the risk type corresponding to the risk type at the comparative risk classification accuracy can be obtained based on the vector matching parameters of the risk types with lower risk classification accuracy and adjacent risk classification accuracy of the risk type; for example, the vector matching parameters of the risk type of the ath risk classification accuracy can be directly determined as the vector matching parameters of the risk type of the comparative risk classification accuracy. For example, the vector matching parameters corresponding to "short-term overdue default" can be directly determined as "extremely low overdue degree", "low overdue degree" and "low overdue degree" of the comparative risk classification accuracy, and the vector matching parameters of the three risk types are the same;

[0112] Finally, the second learning error corresponding to the a-th risk classification accuracy can be determined based on the vector matching parameters of each risk type of the a+1-th risk classification accuracy and the vector matching parameters of the risk types corresponding to the comparative risk classification accuracy; exemplarily, the vector matching parameters of the risk types corresponding to the comparative risk classification accuracy can be normalized first, and then, for the normalized vector matching parameters of the same risk type and the vector matching parameters of the a+1-th risk classification accuracy, the ratio of the latter to the former can be calculated first, and the logarithm of the ratio can be performed to obtain the corresponding logarithm result, and then the product of the latter and the logarithm result can be calculated, and then the product can be inverted to obtain the local second learning error, and finally, the local second learning errors of each risk type can be summed to obtain the second learning error.

[0113] It should be noted that for step S140 , the specific method of obtaining the target risk type corresponding to the target automobile financial risk data is not limited and can be selected according to actual needs.

[0114] For example, in an alternative implementation, in order to determine a reliable target risk category, the above step S140 may further include the following contents:

[0115] First, the risk type corresponding to the risk type description vector with the maximum value of the vector matching parameter can be determined, and the risk type corresponding to the risk type with a lower risk classification accuracy can be determined. For example, the risk type determined in the third risk classification accuracy can be "high degree of overdue", and the risk types with a lower risk classification accuracy corresponding to the risk type can be "long-term overdue default" in the second risk classification accuracy and "will default" in the first risk classification accuracy;

[0116] Secondly, the risk category corresponding to the risk category description vector having the maximum value of the vector matching parameter and the risk category corresponding to the risk category with a lower risk classification accuracy may be marked to form the target risk category corresponding to the target automobile financial risk data.

[0117] In summary, the automobile financial risk assessment method and system based on artificial intelligence provided by the present application first outputs the financial risk description vector corresponding to the target automobile financial risk data; secondly, the risk type, the enhanced description data corresponding to the risk type and the enhanced description data of the risk type with lower risk classification accuracy corresponding to the risk type are merged to form risk type description data; then, the risk type description vector corresponding to the risk type description data is output, and the vector matching parameters respectively possessed by the financial risk description vector and the multiple risk type description vectors are analyzed; finally, the target risk type is obtained according to the risk type corresponding to the risk type description vector with the maximum value of the vector matching parameter. Based on the above content, since the risk type description data includes not only the corresponding risk type, but also the enhanced description data corresponding to the risk type, and also includes the enhanced description data of the risk type with lower risk classification accuracy corresponding to the risk type, the semantic information of the mined risk type description vector can be enriched, thereby providing a more reliable basis for subsequent analysis, so that the reliability of the determined risk type can also be higher, and therefore, the problem of relatively low reliability and efficiency of automobile financial risk assessment existing in the prior art can be improved.

[0118] In several embodiments provided in the embodiments of the present application, it should be understood that the disclosed device and method can also be implemented in other ways. The device and method embodiments described above are merely schematic, for example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the device, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0119] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0120] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, an electronic device, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk. It should be noted that, in this article, the term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such a process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0121] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An automobile financial risk assessment method based on artificial intelligence, characterized in that: include: Performing semantic mining on the target automobile financial risk data and outputting a financial risk description vector corresponding to the target automobile financial risk data, wherein the target automobile financial risk data is used to describe the automobile loan risk situation of the target automobile financial user, and the target automobile financial risk data at least includes basic user information and loan-related information of the target automobile financial user; For each risk type among the multiple risk types with the highest risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with lower risk classification accuracy are merged to form risk type description data, wherein the enhanced description data is formed by learning the sample data, and the risk type with high risk classification accuracy and the corresponding risk type with low risk classification accuracy belong to the lower classification and the upper classification respectively; Performing semantic mining on the risk type description data, outputting a risk type description vector corresponding to the risk type description data, and analyzing vector matching parameters respectively between the financial risk description vector and a plurality of risk type description vectors; The target risk type corresponding to the target automobile financial risk data is obtained based on the risk type corresponding to the risk type description vector whose vector matching parameter has the maximum value among the multiple risk type description vectors, wherein the target risk type corresponding to the target automobile financial risk data is used to reflect the risk type of the target automobile financial user for the corresponding automobile loan.

2. The automobile financial risk assessment method based on artificial intelligence according to claim 1 is characterized in that: The step of performing semantic mining on the target automobile financial risk data and outputting a financial risk description vector corresponding to the target automobile financial risk data includes: Perform word segmentation on the target automobile financial risk data to obtain multiple automobile financial risk words; For each of the plurality of automobile financial risk words, performing word embedding processing on the automobile financial risk word to obtain a corresponding risk word embedding vector; Based on the relationship between the multiple automobile financial risk words, a fusion operation is performed on the multiple risk word embedding vectors corresponding to the multiple automobile financial risk words to output a financial risk description vector corresponding to the target automobile financial risk data.

3. The automobile financial risk assessment method based on artificial intelligence according to claim 2 is characterized in that: The step of fusing a plurality of risk word embedding vectors corresponding to the plurality of automobile financial risk words based on the relationship between the plurality of automobile financial risk words and outputting a financial risk description vector corresponding to the target automobile financial risk data comprises: Performing mean superposition calculation on the risk word embedding vector corresponding to each word in the basic user information in the target automobile financial risk data, outputting the basic user information vector corresponding to the target automobile financial risk data, and performing mean superposition calculation on the risk word embedding vector corresponding to each word in the loan-related information in the target automobile financial risk data, outputting the loan-related information vector corresponding to the target automobile financial risk data; Performing association semantic mining on the basic user information vector and the loan related information vector, and outputting a corresponding first association semantic mining vector; Performing multi-level convolution operations on the basic user information vector and the loan-related information vector respectively, and outputting corresponding multiple basic user information convolution vectors and multiple loan-related information convolution vectors, wherein the convolution vector output by the previous level of convolution operation is used as the convolution vector input by the next level of convolution operation, and there is a one-to-one correspondence between the multiple basic user information convolution vectors and the multiple loan-related information convolution vectors; For each basic user information convolution vector among the multiple basic user information convolution vectors, performing association semantic mining on the basic user information convolution vector and the corresponding loan-related information convolution vector, and outputting a corresponding second association semantic mining vector; The first associated semantic mining vector and each of the second associated semantic mining vectors are concatenated to form a financial risk description vector corresponding to the target automobile financial risk data.

4. The automobile financial risk assessment method based on artificial intelligence according to claim 1 is characterized in that: The step of obtaining the target risk category corresponding to the target automobile financial risk data according to the risk category corresponding to the risk category description vector having the maximum vector matching parameter among the plurality of risk category description vectors comprises: Determine the risk category corresponding to the risk category description vector having the maximum value of the vector matching parameter, and determine the risk category corresponding to the risk category with a lower risk classification accuracy; The risk category corresponding to the risk category description vector having the maximum value of the vector matching parameter and the risk category corresponding to the risk category with a lower risk classification accuracy are marked to form a target risk category corresponding to the target automobile financial risk data.

5. The automobile financial risk assessment method based on artificial intelligence according to any one of claims 1 to 4, characterized in that: The risk types have a target number of risk classification accuracies; Among them, for any two adjacent risk classification accuracies among the target number of risk classification accuracies: the a-th risk classification accuracy and the a+1-th risk classification accuracy, the a-th risk classification accuracy corresponds to a first preset number of risk types, the a+1-th risk classification accuracy corresponds to a second preset number of risk types, one or more risk types in the second preset number of risk types have lower risk classification accuracy and risk types with adjacent risk classification accuracies belong to the first preset number of risk types, the first preset number is not greater than the second preset number, and a+1 is less than or equal to the target number; Furthermore, before executing the step of merging, for each of the multiple risk types with the highest risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with a lower risk classification accuracy corresponding to the risk type to form risk type description data, the automobile financial risk assessment method based on artificial intelligence further includes: Performing semantic mining operations on the sample automobile financial risk data, and outputting a sample financial risk description vector corresponding to the sample automobile financial risk data; For each risk classification accuracy of the target number of risk classification accuracies, determining vector matching parameters corresponding to the multiple risk types of the risk classification accuracy respectively according to the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy and the sample financial risk description vector; Determining a first learning error corresponding to the risk classification accuracy according to vector matching parameters respectively corresponding to a plurality of risk types of the risk classification accuracy; For the a+1th risk classification accuracy among the target number of risk classification accuracies, determine the second learning error corresponding to the ath risk classification accuracy according to the vector matching parameters corresponding to the multiple risk types of the a+1th risk classification accuracy and the vector matching parameters corresponding to the risk types of each risk type of the a+1th risk classification accuracy that has a lower risk classification accuracy and adjacent risk classification accuracy; Obtaining a target learning error according to the first learning error and the second learning error; According to the target learning error, the representation parameters of the reinforcement description data corresponding to multiple risk types of each risk classification accuracy among the target number of risk classification accuracies are optimized to form the reinforcement description data after learning, wherein the representation parameters of the initial reinforcement description data are randomly generated.

6. The automobile financial risk assessment method based on artificial intelligence according to claim 5 is characterized in that: The step of determining, for each of the target number of risk classification accuracies, vector matching parameters corresponding to the multiple risk types of the risk classification accuracy based on the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy and the sample financial risk description vector, comprises: For each risk classification accuracy of the target number of risk classification accuracies, according to the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies, mining the sample risk category description vectors corresponding to the multiple risk categories of the risk classification accuracy; Based on the sample risk type description vectors corresponding to the multiple risk types of the risk classification accuracy, the sample financial risk description vectors and preset adjustment parameters, the vector matching parameters corresponding to the multiple risk types of the risk classification accuracy are determined.

7. The automobile financial risk assessment method based on artificial intelligence according to claim 6 is characterized in that: The step of mining sample risk category description vectors corresponding to a plurality of risk categories of each risk classification accuracy among the target number of risk classification accuracies according to the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies comprises: For each risk classification accuracy among the target number of risk classification accuracies, when the risk classification accuracy does not belong to the highest risk classification accuracy among the target number of risk classification accuracies, for each risk type among multiple risk types of the risk classification accuracy, a merging operation is performed on the risk type and the enhanced description data corresponding to the risk type to form corresponding sample risk type description data; A semantic mining operation is performed on the sample risk type description data to output a sample risk type description vector corresponding to the sample risk type description data.

8. The automobile financial risk assessment method based on artificial intelligence according to claim 6 is characterized in that: The step of mining sample risk category description vectors corresponding to a plurality of risk categories of each risk classification accuracy among the target number of risk classification accuracies according to the upper and lower distribution relationship of the risk classification accuracy over the target number of risk classification accuracies comprises: For each risk classification accuracy among the target number of risk classification accuracy, when the risk classification accuracy belongs to the highest risk classification accuracy among the target number of risk classification accuracy, for each risk type among multiple risk types of the risk classification accuracy, the risk type, the enhanced description data corresponding to the risk type, and the enhanced description data corresponding to the risk type with a lower risk classification accuracy of the risk type are merged to form corresponding sample risk type description data; A semantic mining operation is performed on the sample risk type description data to output a sample risk type description vector corresponding to the sample risk type description data.

9. The automobile financial risk assessment method based on artificial intelligence according to claim 5 is characterized in that: The step of determining a second learning error corresponding to the a+1th risk classification accuracy among the target number of risk classification accuracies according to vector matching parameters respectively corresponding to multiple risk types of the a+1th risk classification accuracy and vector matching parameters corresponding to risk types with lower risk classification accuracy and adjacent risk classification accuracy for each risk type of the a+1th risk classification accuracy comprises: Constructing a comparative risk classification accuracy corresponding to the a-th risk classification accuracy, wherein the number of risk types of the comparative risk classification accuracy is the same as the number of risk types of the a+1-th risk classification accuracy, and there is a one-to-one correspondence between the risk types of the comparative risk classification accuracy and the risk types of the a+1-th risk classification accuracy; For each risk type of the a+1th risk classification accuracy, obtain the vector matching parameters of the risk type corresponding to the comparative risk classification accuracy of the risk type according to the vector matching parameters of the risk types with lower risk classification accuracy and adjacent risk classification accuracy of the risk type; A second learning error corresponding to the a-th risk classification accuracy is determined based on the vector matching parameters of each risk type of the a+1-th risk classification accuracy and the vector matching parameters of the risk type corresponding to the comparative risk classification accuracy.

10. An automobile financial risk assessment system based on artificial intelligence, characterized in that: include: Memory for storing computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the automobile financial risk assessment method based on artificial intelligence as described in any one of claims 1-9.

Images