Car insurance buying information processing method

Abstract

The invention discloses a vehicle insurance information processing method, and particularly relates to the technical field of vehicle insurance, and the method comprises the steps: obtaining the face change information, behavior action information and language fluctuation information of a user during each vehicle accident, building a behavior abnormality analysis model, carrying out the analysis of the abnormal behavior of the user during each vehicle accident, and obtaining the abnormal behavior of the user during each vehicle accident; obtaining a model output result, carrying out clustering analysis on the model output result, determining whether the user has a potential insurance fraud risk, further quantifying the fraud risk level of the user, enabling the risk assessment to be more scientific and objective, building an automatic adjustment model when the user has the potential fraud risk, and improving the user experience. According to the method, the calling period of the insurance information of the user is dynamically adjusted, the requirement of manual intervention is reduced, subjective misjudgment in manual screening is avoided, and the system can generate more reasonable electronic insurance policies meeting actual risks.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile insurance underwriting, and more particularly to a method for processing automobile insurance underwriting information. Background Art

[0002] With the rapid growth of car ownership, auto insurance has become a fundamental safeguard for car owners. The processing of auto insurance information involves multiple steps, including collection, storage, processing, and analysis. Effective management of these steps is crucial for insurance companies to provide accurate insurance services, assess risks, and optimize the customer experience.

[0003] The car insurance information system usually stores the user's insurance information at different times. When a user initiates an insurance application, the user's insurance information at different times is usually manually screened, and the selected insurance information is verified. At the same time, an electronic insurance policy is automatically generated based on the verified insurance information. The insurance information stored in the system usually includes the insurance information of the user's vehicle throughout its life cycle. When a user initiates an insurance application, due to the lack of analysis of the fraud risk in the user's accident before retrieving the insurance information, the insurance information screened by the staff based on their own experience may not support the generation of electronic insurance policies, resulting in errors in the calculation of premiums and affecting the user's insurance experience. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method for processing automobile insurance information to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The method for processing automobile insurance information includes the following steps:

[0007] Step S1, obtaining facial change information, behavioral action information, and language fluctuation information of the user each time a vehicle accident occurs;

[0008] Step S2: constructing a behavior abnormality analysis model to analyze the abnormal behavior of the user each time a vehicle accident occurs, and obtaining a model output result, which includes an abnormal behavior coefficient;

[0009] Step S3: Perform cluster analysis on the model output results to determine whether the user has potential insurance fraud risk;

[0010] Step S4: Automatically adjust the retrieval cycle of user insurance information based on the cluster analysis results.

[0011] In a preferred embodiment, in step S1, the facial change information includes the facial expression change abnormality coefficient, the behavior action information includes the body posture coefficient, and the language fluctuation information includes the language fluctuation coefficient;

[0012] The coefficient of facial expression change abnormality, the coefficient of body posture, and the coefficient of language fluctuation are marked as bqx, ztx, and yyx respectively.

[0013] In a preferred embodiment, the logic for obtaining the facial expression abnormality coefficient is as follows:

[0014] Obtain a facial image of the current user at the time of the vehicle accident, establish a two-dimensional coordinate system based on the facial image, and extract the contour coordinates of key facial features of the user from the two-dimensional coordinate system, including but not limited to eyes, mouth, and eyebrows;

[0015] The cosine similarity calculation method is used to match historically similar users who have had the same type of vehicle accident as the current user from the car insurance information system, as follows:

[0016] Calculate the similarity sim between the current user and the historical user's vehicle accidents. The expression is as follows Where A i represents the i-th vehicle accident feature of the current user, B i represents the i-th vehicle accident feature of the historical user;

[0017] Sort historical users by their accident similarity from large to small to obtain a historical user ranking table, and mark the historical user at the top of the ranking table as a historical similar user who has had the same accident type as the current user;

[0018] Obtain the contour coordinates of the key facial features of historically similar users and calculate the abnormal coefficient of facial expression changes. The expression is as follows Where (x1 j , y1 j ) represents the jth group of contour coordinates of the current user's facial key features, (x2 j , y2 j ) represents the jth group of contour coordinates of the key facial features of historically similar users.

[0019] In a preferred embodiment, the logic for obtaining the limb posture coefficients is as follows:

[0020] Obtain the body posture sequence of the current user when the vehicle accident occurs, and mark the body posture sequence of the current user as DZ = {dz t}={dz1,dz2,...,dz T}, where dz tRepresents the position vector of the t-th limb feature point of the current user, t={1,2,...,T}, where T is a positive integer;

[0021] The cosine similarity calculation method is used to match historically similar users who have had the same type of vehicle accident as the current user from the car insurance information system;

[0022] Obtain the body posture sequence of historically similar users when they have a vehicle accident, and mark the body posture sequence of historically similar users as XZ = {xz t}={xz1,xz2,...,xz T}, where xz t Represents the position vector of the t-th limb feature point of the historically similar user;

[0023] Calculate the Euclidean distance JL between each pair of limb feature points t , according to the Euclidean distance JL between each pair of limb feature points t Calculate the limb posture coefficient, the expression is as follows

[0024] In a preferred embodiment, the logic for obtaining the language fluctuation coefficient is as follows:

[0025] Acquire the user's voice data during the incident handling process, use a pitch detection algorithm to detect the pitch in the voice data, compare the pitch at different times with the pitch threshold, count the number of times the pitch exceeds the pitch threshold, and mark the number of times the pitch exceeds the pitch threshold as ygc;

[0026] Calculate the user's average speaking speed ysp, the expression is as follows Where zsz represents the total number of words in the speech data, and zsj represents the duration of the speech data;

[0027] Calculate the language fluctuation coefficient using the following expression: yyx = ygc * ysp.

[0028] In a preferred embodiment, in step S2, a behavior abnormality analysis model is constructed to analyze the abnormal behavior of the user each time a vehicle accident occurs, as follows:

[0029] The facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient are normalized, and a behavior abnormality analysis model is constructed based on the normalized facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient to generate the abnormal behavior coefficient yxw. The model is based on the following formula: Where a1, a2, and a3 represent the preset proportional coefficients of the facial expression change abnormality coefficient, the body posture coefficient, and the language fluctuation coefficient, respectively, and a1, a2, and a3 are all greater than 0.

[0030] In a preferred embodiment, in step S3, cluster analysis is performed on the model output results, as follows:

[0031] Step S31, obtaining analysis results of abnormal behavior of the user each time a vehicle accident occurs, that is, obtaining multiple abnormal behavior coefficients, and using the maximum and minimum values ​​of the abnormal behavior coefficients as initial cluster centers;

[0032] Step S32, calculating the Euclidean distance between each abnormal behavior coefficient and the cluster center, assigning each abnormal behavior coefficient to the cluster center with the closest Euclidean distance, and establishing a standard cluster of abnormal behavior coefficients;

[0033] Step S33, calculating the average value of the abnormal behavior coefficients within each abnormal behavior coefficient standard cluster and using it as the new cluster center;

[0034] Step S34, repeating steps S32 and S33 to iteratively update the cluster center until the cluster center no longer changes;

[0035] Step S35, calculate the fraud risk index qzz, the expression is as follows Where E m represents the abnormal behavior coefficient of the user’s mth vehicle accident output, C k Indicates E m The cluster center of the abnormal behavior coefficient standard cluster k, R k The range of the abnormal behavior coefficient within the abnormal behavior coefficient standard cluster k is obtained by subtracting the minimum value from the maximum value of the abnormal behavior coefficient within the abnormal behavior coefficient standard cluster k. k represents the weight value of the abnormal behavior coefficient standard cluster k, and M represents the number of vehicle accidents that occurred to the user.

[0036] In a preferred embodiment, the fraud risk index is compared with a preset fraud risk index threshold. If the fraud risk index is greater than the fraud risk index threshold, an insurance fraud risk signal is generated; if the fraud risk index is less than or equal to the fraud risk index threshold, no insurance fraud risk signal is required.

[0037] In a preferred embodiment, in step S4, the retrieval period of the user's insurance information is automatically adjusted according to the cluster analysis results, as follows:

[0038] When an insurance fraud risk signal is generated, the corresponding fraud risk index and fraud risk index threshold are obtained and used as adjustment factors for automatically adjusting the retrieval cycle of user insurance information, and an automatic adjustment model is constructed. The model is based on the following formula: Where zq new Indicates a new call cycle, zq max Indicates the maximum time span for retrieving user insurance information, zqmin Indicates the minimum time span for retrieving user insurance information, qzz indicates the fraud risk index, yz Indicates the fraud risk index threshold.

[0039] The technical effects and advantages of the present invention are as follows:

[0040] 1. The present invention constructs a behavioral anomaly analysis model by acquiring facial change information, behavioral action information, and language fluctuation information of the user each time a vehicle accident occurs, analyzes the user's abnormal behavior each time a vehicle accident occurs, obtains the model output results, and performs cluster analysis on the model output results to determine whether the user has potential insurance fraud risks, further quantify the user's fraud risk level, and make risk assessment more scientific and objective. When a user has potential fraud risks, an automated adjustment model is constructed to dynamically adjust the retrieval cycle of the user's insurance information, reduce the need for manual intervention, avoid subjective misjudgments in manual screening, and enable the system to generate more reasonable electronic insurance policies that conform to actual risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0042] Figure 1 Schematic diagram of the structure of the method of the embodiment of the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] Example: Figure 1 The present invention provides a method for processing automobile insurance information, comprising the following steps:

[0045] Step S1, obtaining facial change information, behavioral action information, and language fluctuation information of the user each time a vehicle accident occurs;

[0046] Step S2: constructing a behavior abnormality analysis model to analyze the abnormal behavior of the user each time a vehicle accident occurs, and obtaining a model output result, which includes an abnormal behavior coefficient;

[0047] Step S3: Perform cluster analysis on the model output results to determine whether the user has potential insurance fraud risk;

[0048] Step S4: Automatically adjust the retrieval cycle of user insurance information based on the cluster analysis results;

[0049] In step S1, sensors such as cameras and microphones installed in the vehicle or at the accident scene are used to collect facial change information, behavioral information, and language fluctuation information of the user each time a vehicle accident occurs. The facial change information includes the abnormal coefficient of facial expression changes, the behavioral information includes the body posture coefficient, and the language fluctuation information includes the language fluctuation coefficient.

[0050] The coefficient of facial expression change abnormality, the coefficient of body posture, and the coefficient of language fluctuation are marked as bqx, ztx, and yyx respectively;

[0051] The Facial Expression Abnormality Coefficient (FAC) is an important indicator used to measure whether a user's facial expression changes abnormally during a vehicle accident. Specifically, it reflects the degree to which the user's facial expression deviates from normal at the moment of the accident and during the accident handling process. The magnitude of this coefficient can be used to identify abnormal user emotional reactions, such as anxiety, tension, anger, or excessive calmness. A larger FAC indicates an abnormal user emotional reaction and may indicate fraudulent behavior.

[0052] The logic for obtaining the abnormal coefficient of facial expression changes is as follows:

[0053] Obtain a facial image of the current user at the time of the vehicle accident, establish a two-dimensional coordinate system based on the facial image, and extract the contour coordinates of key facial features of the user from the two-dimensional coordinate system, including but not limited to eyes, mouth, and eyebrows;

[0054] The cosine similarity calculation method is used to match historically similar users who have had the same type of vehicle accident as the current user from the car insurance information system, as follows:

[0055] Calculate the similarity sim between the current user and the historical user's vehicle accidents. The expression is as follows Where A i represents the i-th vehicle accident feature of the current user, B i represents the i-th vehicle accident feature of the historical user;

[0056] It should be noted that the characteristics of vehicle accidents include the damaged part, damaged area, number of passengers, duration of the accident, etc.

[0057] The vehicle accident similarity is used to measure the similarity between the types of vehicle accidents that occurred to the current user and those of previous users. The greater the vehicle accident similarity, the higher the similarity between the types of vehicle accidents that occurred to the current user and those of previous users. Conversely, the smaller the vehicle accident similarity, the lower the similarity between the types of vehicle accidents that occurred to the current user and those of previous users.

[0058] Sort historical users by their accident similarity from large to small to obtain a historical user ranking table, and mark the historical user at the top of the ranking table as a historical similar user who has had the same accident type as the current user;

[0059] Obtain the contour coordinates of the key facial features of historically similar users and calculate the abnormal coefficient of facial expression changes. The expression is as follows Where (x1 j , y1 j ) represents the jth group of contour coordinates of the current user's facial key features, (x2 j , y2 j ) represents the jth group of contour coordinates of the key facial features of historically similar users;

[0060] The body posture coefficient is a key indicator used to measure whether a user's body movements are abnormal during a vehicle accident. Specifically, it reflects the degree to which the user's body movements deviate from the normal state at the moment of the accident and during the accident handling process. The size of this coefficient can be used to identify whether the user's body reactions are abnormal, such as anxiety, tension, avoidance, unnatural movements, and other behavioral manifestations. A larger body posture coefficient indicates that the user's body movements are abnormal, which may indicate abnormal behavior and may lead to suspicion of fraud.

[0061] The logic for obtaining limb posture coefficients is as follows:

[0062] Obtain the body posture sequence of the current user when the vehicle accident occurs, and mark the body posture sequence of the current user as DZ = {dz t}={dz1,dz2,...,dz T}, where dz t Represents the position vector of the t-th limb feature point of the current user, t={1,2,...,T}, where T is a positive integer;

[0063] It should be noted that limb feature points include shoulders, elbows, wrists, palms, knees, ankles, toes, etc.

[0064] The cosine similarity calculation method is used to match historically similar users who have had the same type of vehicle accident as the current user from the car insurance information system;

[0065] It should be noted that the method for obtaining historically similar users is the same as above and will not be elaborated here;

[0066] Obtain the body posture sequence of historically similar users when they have a vehicle accident, and mark the body posture sequence of historically similar users as XZ = {xz t}={xz1,xz2,...,xz T}, where xz t Represents the position vector of the t-th limb feature point of the historically similar user;

[0067] Calculate the Euclidean distance JL between each pair of limb feature points t , according to the Euclidean distance JL between each pair of limb feature points t Calculate the limb posture coefficient, the expression is as follows

[0068] The Language Fluctuation Coefficient is an important indicator used to measure whether a user's language expression is abnormal after a vehicle accident. It reflects any abnormalities in the user's language intonation, speaking speed, and speech frequency during the post-accident process. The magnitude of this coefficient can help identify whether a user is displaying unusual emotions or behaviors. A large Language Fluctuation Coefficient indicates abnormal language expression and may indicate fraudulent behavior.

[0069] The logic for obtaining the language fluctuation coefficient is as follows:

[0070] Acquire the user's voice data during the incident handling process, use a pitch detection algorithm to detect the pitch in the voice data, compare the pitch at different times with the pitch threshold, count the number of times the pitch exceeds the pitch threshold, and mark the number of times the pitch exceeds the pitch threshold as ygc;

[0071] Calculate the user's average speaking speed ysp, the expression is as follows Where zsz represents the total number of words in the speech data, and zsj represents the duration of the speech data;

[0072] Calculate the language fluctuation coefficient, the expression is as follows yyx = ygc * ysp;

[0073] In step S2, a behavior anomaly analysis model is constructed to analyze the abnormal behavior of the user each time a vehicle accident occurs, as follows:

[0074] The facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient are normalized, and a behavior abnormality analysis model is constructed based on the normalized facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient to generate the abnormal behavior coefficient yxw. The model is based on the following formula: Where a1, a2, and a3 represent the preset proportional coefficients of the facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient, respectively, and a1, a2, and a3 are all greater than 0;

[0075] From the above calculation expression, it can be seen that the larger the coefficient of abnormal facial expression change, the larger the coefficient of body posture, and the larger the coefficient of language fluctuation, the larger the abnormal behavior coefficient, indicating that the probability that the user's behavior may be abnormal each time a vehicle accident occurs is higher. Conversely, the smaller the coefficient of abnormal facial expression change, the smaller the coefficient of body posture, and the smaller the coefficient of language fluctuation, the smaller the abnormal behavior coefficient, indicating that the probability that the user's behavior may be abnormal each time a vehicle accident occurs is lower.

[0076] In step S3, cluster analysis is performed on the model output results, as follows:

[0077] Step S31, obtaining analysis results of abnormal behavior of the user each time a vehicle accident occurs, that is, obtaining multiple abnormal behavior coefficients, and using the maximum and minimum values ​​of the abnormal behavior coefficients as initial cluster centers;

[0078] Step S32, calculating the Euclidean distance between each abnormal behavior coefficient and the cluster center, assigning each abnormal behavior coefficient to the cluster center with the closest Euclidean distance, and establishing a standard cluster of abnormal behavior coefficients;

[0079] Step S33, calculating the average value of the abnormal behavior coefficients within each abnormal behavior coefficient standard cluster and using it as the new cluster center;

[0080] Step S34, repeating steps S32 and S33 to iteratively update the cluster center until the cluster center no longer changes;

[0081] Step S35, calculate the fraud risk index qzz, the expression is as follows Where E m represents the abnormal behavior coefficient of the user’s mth vehicle accident output, C k Indicates E m The cluster center of the abnormal behavior coefficient standard cluster k, R k The range of the abnormal behavior coefficient within the abnormal behavior coefficient standard cluster k is obtained by subtracting the minimum value from the maximum value of the abnormal behavior coefficient within the abnormal behavior coefficient standard cluster k. k represents the weight value of the abnormal behavior coefficient standard cluster k, and M represents the number of vehicle accidents that the user has experienced;

[0082] The fraud risk index is used to measure whether a user has a potential risk of insurance fraud. The larger the fraud risk index, the higher the probability that the user has a potential risk of insurance fraud. Conversely, the smaller the fraud risk index, the lower the probability that the user has a potential risk of insurance fraud.

[0083] Compare the fraud risk index with the preset fraud risk index threshold. If the fraud risk index is greater than the fraud risk index threshold, a fraud risk signal is generated, and the retrieval cycle of the user's insurance information is automatically adjusted. If the fraud risk index is less than or equal to the fraud risk index threshold, no fraud risk signal is required, and the user's insurance information can be retrieved according to the retrieval rules set by the system.

[0084] In step S4, based on the cluster analysis results, the retrieval cycle of the user's insurance information is automatically adjusted as follows:

[0085] When an insurance fraud risk signal is generated, the corresponding fraud risk index and fraud risk index threshold are obtained and used as adjustment factors for automatically adjusting the retrieval cycle of user insurance information, and an automatic adjustment model is constructed. The model is based on the following formula: Where zq new Indicates a new call cycle, zq max Indicates the maximum time span for retrieving user insurance information, zq min Indicates the minimum time span for retrieving user insurance information, qzz indicates the fraud risk index, yz represents the fraud risk index threshold;

[0086] Put the new retrieval cycle obtained into practical use and automatically adjust the retrieval cycle of user insurance information;

[0087] The present invention obtains the user's facial change information, behavioral action information and language fluctuation information each time a vehicle accident occurs, constructs a behavioral anomaly analysis model, analyzes the user's abnormal behavior each time a vehicle accident occurs, obtains the model output results, and performs cluster analysis on the model output results to determine whether the user has potential insurance fraud risks, further quantify the user's fraud risk level, and make risk assessment more scientific and objective. When a user has potential fraud risks, an automated adjustment model is constructed to dynamically adjust the retrieval cycle of the user's insurance information, reduce the need for manual intervention, avoid subjective misjudgments in manual screening, and enable the system to generate more reasonable electronic insurance policies that conform to actual risks.

[0088] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0089] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not 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 embodiments of the present application.

[0090] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for processing automobile insurance information, characterized by: The steps include: Step S1, obtaining facial change information, behavioral action information, and language fluctuation information of the user each time a vehicle accident occurs; Step S2: constructing a behavior abnormality analysis model to analyze the abnormal behavior of the user each time a vehicle accident occurs, and obtaining a model output result, which includes an abnormal behavior coefficient; Step S3: Perform cluster analysis on the model output results to determine whether the user has potential insurance fraud risk; Step S4: Automatically adjust the retrieval cycle of user insurance information based on the cluster analysis results.

2. The method for processing automobile insurance information according to claim 1, characterized in that: In step S1, the facial change information includes the facial expression change abnormality coefficient, the behavior action information includes the body posture coefficient, and the language fluctuation information includes the language fluctuation coefficient; The coefficient of facial expression change abnormality, the coefficient of body posture, and the coefficient of language fluctuation are marked as bqx, ztx, and yyx respectively.

3. The method for processing automobile insurance information according to claim 2, characterized in that: The logic for obtaining the abnormal coefficient of facial expression changes is as follows: Obtain a facial image of the current user at the time of the vehicle accident, establish a two-dimensional coordinate system based on the facial image, and extract the contour coordinates of key facial features of the user from the two-dimensional coordinate system, including but not limited to eyes, mouth, and eyebrows; The cosine similarity calculation method is used to match historically similar users who have had the same type of vehicle accident as the current user from the car insurance information system, as follows: Calculate the similarity sim between the current user and the historical user's vehicle accidents. The expression is as follows Where A i represents the i-th vehicle accident feature of the current user, B i represents the i-th vehicle accident feature of the historical user; Sort historical users by their accident similarity from large to small to obtain a historical user ranking table, and mark the historical user at the top of the ranking table as a historical similar user who has had the same accident type as the current user; Obtain the contour coordinates of the key facial features of historically similar users and calculate the abnormal coefficient of facial expression changes. The expression is as follows Where (x1 j , y1 j ) represents the jth group of contour coordinates of the current user's facial key features, (x2 j , y2 j ) represents the jth group of contour coordinates of the key facial features of historically similar users.

4. The method for processing automobile insurance information according to claim 2, characterized in that: The logic for obtaining limb posture coefficients is as follows: Obtain the body posture sequence of the current user when the vehicle accident occurs, and mark the body posture sequence of the current user as DZ = {dz t }={dz1,dz2,...,dz T }, where dz t Represents the position vector of the t-th limb feature point of the current user, t={1,2,...,T}, where T is a positive integer; The cosine similarity calculation method is used to match historically similar users who have had the same type of vehicle accident as the current user from the car insurance information system; Obtain the body posture sequence of historically similar users when they have a vehicle accident, and mark the body posture sequence of historically similar users as XZ = {xz t }={xz1,xz2,...,xz T }, where xz t Represents the position vector of the t-th limb feature point of the historically similar user; Calculate the Euclidean distance JL between each pair of limb feature points t , according to the Euclidean distance JL between each pair of limb feature points t Calculate the limb posture coefficient, the expression is as follows 5. The method for processing automobile insurance information according to claim 2, characterized in that: The logic for obtaining the language fluctuation coefficient is as follows: Acquire the user's voice data during the incident handling process, use a pitch detection algorithm to detect the pitch in the voice data, compare the pitch at different times with the pitch threshold, count the number of times the pitch exceeds the pitch threshold, and mark the number of times the pitch exceeds the pitch threshold as ygc; Calculate the user's average speaking speed ysp, the expression is as follows Where zsz represents the total number of words in the speech data, and zsj represents the duration of the speech data; Calculate the language fluctuation coefficient using the following expression: yyx = ygc * ysp.

6. The method for processing automobile insurance information according to claim 5, characterized in that: In step S2, a behavior anomaly analysis model is constructed to analyze the abnormal behavior of the user each time a vehicle accident occurs, as follows: The facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient are normalized, and a behavior abnormality analysis model is constructed based on the normalized facial expression change abnormality coefficient, body posture coefficient, and language fluctuation coefficient to generate the abnormal behavior coefficient yxw. The model is based on the following formula: Where a1, a2, and a3 represent the preset proportional coefficients of the facial expression change abnormality coefficient, the body posture coefficient, and the language fluctuation coefficient, respectively, and a1, a2, and a3 are all greater than 0.

7. The method for processing automobile insurance information according to claim 6, characterized in that: In step S3, cluster analysis is performed on the model output results, as follows: Step S31, obtaining analysis results of abnormal behavior of the user each time a vehicle accident occurs, that is, obtaining multiple abnormal behavior coefficients, and using the maximum and minimum values ​​of the abnormal behavior coefficients as initial cluster centers; Step S32, calculating the Euclidean distance between each abnormal behavior coefficient and the cluster center, assigning each abnormal behavior coefficient to the cluster center with the closest Euclidean distance, and establishing a standard cluster of abnormal behavior coefficients; Step S33, calculating the average value of the abnormal behavior coefficients within each abnormal behavior coefficient standard cluster and using it as the new cluster center; Step S34, repeating steps S32 and S33 to iteratively update the cluster center until the cluster center no longer changes; Step S35, calculate the fraud risk index qzz, the expression is as follows Where E m represents the abnormal behavior coefficient of the user’s mth vehicle accident output, C k Indicates E m The cluster center of the abnormal behavior coefficient standard cluster k, R k The range of the abnormal behavior coefficient within the abnormal behavior coefficient standard cluster k is obtained by subtracting the minimum value from the maximum value of the abnormal behavior coefficient within the abnormal behavior coefficient standard cluster k. k represents the weight value of the abnormal behavior coefficient standard cluster k, and M represents the number of vehicle accidents that occurred to the user.

8. The method for processing automobile insurance information according to claim 7, characterized in that: Compare the fraud risk index with a preset fraud risk index threshold; if the fraud risk index is greater than the fraud risk index threshold, generate an insurance fraud risk signal; If the fraud risk index is less than or equal to the fraud risk index threshold, there is no need for an insurance fraud risk signal.

9. The method for processing automobile insurance information according to claim 8, characterized in that: In step S4, based on the cluster analysis results, the retrieval cycle of the user's insurance information is automatically adjusted as follows: When an insurance fraud risk signal is generated, the corresponding fraud risk index and fraud risk index threshold are obtained and used as adjustment factors for automatically adjusting the retrieval cycle of user insurance information, and an automatic adjustment model is constructed. The model is based on the following formula: Where zq new Indicates a new call cycle, zq max Indicates the maximum time span for retrieving user insurance information, zq min Indicates the minimum time span for retrieving user insurance information, qzz indicates the fraud risk index, yz Indicates the fraud risk index threshold.

Images