Business risk processing method based on artificial intelligence

Abstract

The invention discloses a business risk processing method based on artificial intelligence, comprising target user identification and feature extraction, the target user identification and feature extraction comprises elderly user preliminary determination and feature extraction, the elderly user identification and feature extraction can accurately identify an elderly user group, and the elderly user group can be accurately identified through the elderly user identification and feature extraction. According to the method and the system provided by the invention, the risk behavior is accurately identified according to unique operation habits, transaction preferences and other characteristics, so that the ability of identifying fraud risks possibly faced by the old people is greatly improved, the risk that the old people are defrauded is effectively reduced, and abnormal or suspicious transaction behaviors can be rapidly identified through real-time analysis of transaction information, thereby improving the safety of the old people. And an early warning mechanism is triggered in time, so that intervention measures can be taken in time before the risk occurs, the property safety of the elderly can be protected, and meanwhile, the result can be identified according to the personal characteristics of the elderly user and the risk.

Description

Technical Field

[0001] The present invention relates to the technical field of methods for handling business risks of bank automatic teller machines for the elderly, and in particular to a business risk handling method based on artificial intelligence. Background Art

[0002] With the advent of an aging society, the elderly have an increasing demand for cash withdrawals at bank ATMs. However, due to their special physiological and psychological characteristics, the diversity of fraud methods, and their relatively low ability to apply technology, the elderly are easily targeted by fraudsters when using ATMs. In particular, when fraudsters use technical means to forge false interfaces or send fraudulent information, the elderly often find it difficult to distinguish and are deceived into transferring large amounts of cash;

[0003] Although existing bank ATMs have some basic security measures, such as password input protection and transaction record query, they are still insufficient in risk prevention for the elderly. For example, there is a lack of risk warning mechanism for the elderly, and it is impossible to provide personalized risk warnings based on the elderly's operating habits, transaction records and other information, resulting in the elderly lacking sufficient awareness of prevention when facing risks. Therefore, a business risk management method based on artificial intelligence is proposed to address the above situation. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings that the existing technology still has deficiencies in risk prevention for the elderly population, such as the lack of a risk warning mechanism for the elderly population, and the inability to provide personalized risk warnings based on the elderly's operating habits, transaction records and other information, resulting in the elderly lacking sufficient awareness of prevention when facing risks. A business risk management method based on artificial intelligence is proposed to solve the problems.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A business risk processing method based on artificial intelligence, comprising:

[0007] Identification and feature extraction of target users, the identification and feature extraction of target users includes:

[0008] S1: Preliminary judgment of elderly users

[0009] Obtain the images captured by the camera and perform preliminary identification of the target object in the image through machine learning. If the object is preliminarily identified as an elderly person, perform specific feature verification.

[0010] Analyze the captured image and determine the age of the target subject through the recognition module.

[0011] S2: Feature extraction

[0012] The age of the subject is determined based on the extracted shooting data containing pictures of human features, and is determined based on facial features, behavioral features and transaction features.

[0013] Identify whether the target object is an elderly person. Once identified as an elderly person, the system automatically enters the elderly mode and extracts information such as the user's facial features and behavioral characteristics

[0014] It also includes risk behavior identification, which is a risk behavior identification model built based on a deep learning algorithm.

[0015] It also includes risky transaction identification and early warning, which includes:

[0016] S3: Real-time transaction monitoring

[0017] The system captures and records the target user's transaction data in real time, including key information such as withdrawal amount, transaction time, and account balance.

[0018] S4: Transaction Risk Assessment

[0019] Leverage machine learning algorithms to conduct risk assessment on transaction data captured in real time.

[0020] S5: Risk early warning mechanism

[0021] When the system identifies a high-risk transaction, it will immediately trigger an early warning mechanism, which includes voice prompts, SMS notifications, and notifications to bank staff.

[0022] S6: Risk treatment and recording

[0023] Bank staff will intervene in a timely manner and the system will record the processing process and results of each risk warning to provide a basis for subsequent risk analysis and improvements.

[0024] Preferably, the extracted features mainly include:

[0025] Facial features: Through face recognition technology, facial features of elderly users, such as wrinkles and skin sagging, are extracted.

[0026] Behavioral characteristics: Through cameras and sensors, the behavioral patterns of elderly users when using ATMs, such as operation speed, key sequence, and body posture, are captured.

[0027] Transaction characteristics: Record the transaction history of elderly users, including withdrawal amount, transaction time, account balance and other information.

[0028] Preferably, the steps of constructing the risk behavior identification model are:

[0029] S001: The system captures the elderly’s withdrawal behavior data in real time through the optical camera on the ATM.

[0030] S002: The system uses a pre-trained deep learning model to analyze these behavioral data. The model is trained with a large amount of historical data and machine learning algorithms, and can identify abnormal behaviors that do not conform to normal behavioral patterns.

[0031] S003: The system conducts a comprehensive analysis based on the elderly’s historical transaction records, account balances and other information.

[0032] Preferably, the transaction risk assessment method for determining real-time transaction monitoring data is:

[0033] Withdrawal amount: If the withdrawal amount exceeds a certain percentage of the elderly person’s daily transaction habits or account balance, it will be considered a high-risk transaction.

[0034] Transaction time: Withdrawal operations performed during non-regular trading hours will increase the risk of being defrauded and will be considered a high-risk transaction.

[0035] Transaction frequency: Frequent withdrawals within a short period of time indicate abnormal activity in the account and will be considered high-risk transactions.

[0036] Account balance: If the account balance after withdrawal is too low, resulting in the elderly being unable to live a normal life or cope with emergencies, it will be considered a high-risk transaction.

[0037] Preferably, the early warning steps of the risk early warning mechanism are:

[0038] S004: Through the voice system of the ATM, a warning sound is issued to the elderly to remind them to pay attention to transaction safety.

[0039] S005: Send an early warning message to the elderly person’s registered mobile phone number to remind them to verify the transaction.

[0040] S006: The system will also push the warning information to the bank's back-end management system for further verification and processing by staff.

[0041] Preferably, it also includes personalized risk warnings and operation guidance, and the operation steps of the personalized risk warnings and operation guidance are:

[0042] S007: The system constructs a personalized risk profile based on the target user characteristics, transaction records and risk identification results collected in the previous steps.

[0043] S008: Generate customized risk alerts for each target user based on the risk profile.

[0044] S009: Provide personalized operational guidance and continuously optimize the content of risk warnings and operational guidance based on user feedback and real-time risk identification results.

[0045] Preferably, it also includes risk handling process optimization and recording, and the risk handling process optimization includes optimizing the risk handling process and recording the process and results of each risk handling.

[0046] Preferably, the steps of risk management recording are:

[0047] S010: Risk identification records

[0048] Record risk sources, risk events, potential impacts and other information in detail.

[0049] S011: Risk Assessment Records

[0050] Conduct qualitative and quantitative assessment of risks, and record the assessment results, basis and process.

[0051] S012: Risk management records

[0052] Record risk management measures, responsible persons, implementation time and other information.

[0053] S013: Risk monitoring records

[0054] Regularly record information such as changes in risks and effects of treatment.

[0055] Preferably, it also includes fraud information identification and interception:

[0056] Through natural language processing and image recognition technology, the system monitors in real time what is displayed on the ATM screen and the conversations that elderly users may be exchanging with fraudsters via their mobile phones. If fraudulent information is identified, such as false transfer prompts, fake bank interfaces, etc., the system immediately intercepts and closes related operations, records the content of the conversation, and sends alerts to the elderly and the bank.

[0057] The steps to identify and block fraudulent information are as follows:

[0058] S014: Information acquisition and preprocessing

[0059] The system collects and monitors all interactive information of bank ATMs in real time, including screen display content, key input, and voice communication of elderly users.

[0060] The collected information is preprocessed to remove noise and irrelevant content and extract key information such as transaction amount, recipient account and transfer notes.

[0061] S015: Feature extraction and model matching

[0062] Machine learning algorithms are used to extract features from preprocessed information and identify characteristic patterns related to fraudulent behavior.

[0063] The extracted features are matched with the pre-trained fraud information recognition model to determine whether there is a fraud risk.

[0064] S016: Real-time risk assessment

[0065] Combine the target user's personal characteristics and current transaction information to conduct real-time risk assessment.

[0066] If the assessment results determine that there is a high risk of fraud, the system will immediately trigger an early warning mechanism.

[0067] S017: Fraud information interception:

[0068] Once an early warning is triggered, the system will automatically intercept transactions suspected of fraud and suspend fund transfer operations. The system will inform elderly users through voice prompts, screen displays, etc. that the transaction has been intercepted and remind users to be vigilant against fraud.

[0069] S018: Subsequent processing and feedback:

[0070] The system will record the intercepted fraud information and related data to provide data support for subsequent risk analysis and model optimization. At the same time, the system will feedback the interception results to users through text messages, phone calls, etc., and guide users to perform subsequent operations, such as re-initiating transactions or reporting alarms.

[0071] S019: Model update and optimization:

[0072] Based on the continuously collected fraud information and user feedback, the fraud information identification model is regularly updated and optimized to improve the identification accuracy and interception efficiency.

[0073] The present invention has the following beneficial effects:

[0074] 1. The present invention can accurately identify the elderly user group through elderly user identification and feature extraction, and accurately identify risky behaviors based on their unique operating habits, transaction preferences and other characteristics. This greatly improves the ability to identify fraud risks that the elderly may face and effectively reduces the risk of the elderly being defrauded of their funds.

[0075] 2. The present invention can quickly identify abnormal or suspicious transaction behaviors and trigger the early warning mechanism in time through real-time analysis of transaction information, which helps to take timely intervention measures before risks occur to protect the property safety of the elderly. At the same time, according to the personal characteristics of elderly users and risk identification results, the present invention can generate customized risk warnings and operational guidance. This personalized service not only improves the elderly's awareness of risks, but also helps them better understand and deal with possible risks through concise and clear prompts and guidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 It is a schematic diagram of the overall process proposed by the present invention;

[0077] Figure 2 This is a schematic diagram of the elderly user identification and feature extraction process proposed by the present invention;

[0078] Figure 3 This is a schematic diagram of the risk behavior identification process proposed by the present invention;

[0079] Figure 4 This is a schematic diagram of the risk transaction identification and early warning process proposed by the present invention;

[0080] Figure 5 A schematic diagram of the personalized risk warning and operation guidance process proposed by the present invention;

[0081] Figure 6 This is a schematic diagram of the risk management process optimization and recording process proposed in the present invention. DETAILED DESCRIPTION

[0082] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0083] Reference Figure 1-6 , an artificial intelligence-based business risk management method, including

[0084] Identification and feature extraction of target users, the identification and feature extraction of target users includes:

[0085] S1: Preliminary judgment of elderly users

[0086] Obtain the images captured by the camera and perform preliminary identification of the target object in the image through machine learning. If the object is preliminarily identified as an elderly person, perform specific feature verification.

[0087] Analyze the captured image and determine the age of the target subject through the recognition module.

[0088] S2: Feature extraction

[0089] The age of the subject is determined based on the extracted shooting data containing pictures of human features, and is determined based on facial features, behavioral features and transaction features.

[0090] Identify whether the target object is an elderly person. Once identified as an elderly person, the system automatically enters the elderly mode and extracts information such as the user's facial features and behavioral characteristics

[0091] It also includes risk behavior identification, which is a risk behavior identification model built based on a deep learning algorithm.

[0092] It also includes risky transaction identification and early warning, which includes:

[0093] S3: Real-time transaction monitoring

[0094] The system captures and records the target user's transaction data in real time, including key information such as withdrawal amount, transaction time, and account balance.

[0095] S4: Transaction Risk Assessment

[0096] Leverage machine learning algorithms to conduct risk assessment on transaction data captured in real time.

[0097] S5: Risk early warning mechanism

[0098] When the system identifies a high-risk transaction, it will immediately trigger an early warning mechanism, which includes voice prompts, SMS notifications, and notifications to bank staff.

[0099] S6: Risk treatment and recording

[0100] Bank staff will intervene in a timely manner and the system will record the processing process and results of each risk warning to provide a basis for subsequent risk analysis and improvements.

[0101] Specifically, the extracted features mainly include:

[0102] Facial features: Through face recognition technology, facial features of elderly users, such as wrinkles and skin sagging, are extracted.

[0103] Behavioral characteristics: Through cameras and sensors, the behavioral patterns of elderly users when using ATMs, such as operation speed, key sequence, and body posture, are captured.

[0104] Transaction characteristics: Record the transaction history of elderly users, including withdrawal amount, transaction time, account balance and other information.

[0105] Specifically, the steps for constructing the risk behavior identification model are:

[0106] S001: The system captures the elderly’s withdrawal behavior data in real time through the optical camera on the ATM.

[0107] S002: The system uses a pre-trained deep learning model to analyze these behavioral data. The model is trained with a large amount of historical data and machine learning algorithms, and can identify abnormal behaviors that do not conform to normal behavioral patterns.

[0108] S003: The system conducts a comprehensive analysis based on the elderly’s historical transaction records, account balances and other information.

[0109] Specifically, the transaction risk assessment method for determining real-time transaction monitoring data is as follows:

[0110] Withdrawal amount: If the withdrawal amount exceeds a certain percentage of the elderly person’s daily transaction habits or account balance, it will be considered a high-risk transaction.

[0111] Transaction time: Withdrawal operations performed during non-regular trading hours will increase the risk of being defrauded and will be considered a high-risk transaction.

[0112] Transaction frequency: Frequent withdrawals within a short period of time indicate abnormal activity in the account and will be considered high-risk transactions.

[0113] Account balance: If the account balance after withdrawal is too low, resulting in the elderly being unable to live a normal life or cope with emergencies, it will be considered a high-risk transaction.

[0114] Specifically, the early warning steps of the risk early warning mechanism are:

[0115] S004: Through the voice system of the ATM, a warning sound is issued to the elderly to remind them to pay attention to transaction safety.

[0116] S005: Send an early warning message to the elderly person’s registered mobile phone number to remind them to verify the transaction.

[0117] S006: The system will also push the warning information to the bank's back-end management system for further verification and processing by staff.

[0118] The early warning information will record in detail the type, time, location and risk level of the identified risky behavior, and remind the elderly and bank staff through voice prompts, text message notifications, etc. Relevant personnel will be able to respond quickly and take appropriate risk control measures to protect the property safety of the elderly.

[0119] Specifically, it also includes personalized risk warnings and operation guidance, and the operation steps of the personalized risk warnings and operation guidance are:

[0120] S007: The system constructs a personalized risk profile based on the target user characteristics, transaction records and risk identification results collected in the previous steps. The profile will record in detail the user's age, gender, operating habits, transaction preferences and possible risk types.

[0121] S008: Generate customized risk warnings for each target user based on the risk profile. For example, for elderly users who frequently make large withdrawals, the system will remind them to be careful about fraud risks and avoid leaking personal information or bank card passwords in public places. For users who operate slowly or do not press keys in a standard sequence, the system will remind them to pay attention to operating specifications to avoid risks caused by accidental touches or operational errors.

[0122] S009: Provide personalized operation guidance and continuously optimize the content of risk warnings and operation guidance based on user feedback and real-time risk identification results. These guidance contents will target the operation habits and possible risk points of elderly users, and guide users to complete the withdrawal operation correctly and safely through concise and clear text and image presentations. For example, the system can guide users on how to correctly place bank cards, how to enter passwords, how to check transaction information, etc., to ensure that each step of the operation complies with safety regulations. If users encounter confusion or problems during the operation, the system can provide answers and help in a timely manner to ensure that they can successfully complete the withdrawal operation.

[0123] Through the implementation of personalized risk warnings and operational guidance, a more considerate and professional service experience can be provided to elderly users. This will not only help improve their confidence and satisfaction in using bank ATMs, but also effectively reduce the risks they face during the withdrawal process and protect their property safety.

[0124] Specifically, it also includes risk handling process optimization and recording. The risk handling process optimization includes optimizing the risk handling process and recording the process and results of each risk handling. When optimizing the risk handling process, the existing process is comprehensively reviewed and analyzed to identify existing bottlenecks, deficiencies or redundant links. Then, based on business needs and risk management goals, a more reasonable and efficient processing process is designed to ensure that risks are identified, evaluated and handled in a timely and accurate manner. During the optimization process, a series of methods and techniques can be used, such as flowchart analysis, critical path method, Six Sigma management, etc., to improve the efficiency and quality of the process.

[0125] Specifically, the steps of risk management record are:

[0126] S010: Risk identification records

[0127] Record risk sources, risk events, potential impacts and other information in detail to ensure that risks are identified comprehensively and accurately.

[0128] S011: Risk Assessment Records

[0129] Conduct qualitative and quantitative assessments of risks, record assessment results, basis and process, and provide a basis for risk management.

[0130] S012: Risk management records

[0131] Record risk management measures, responsible persons, implementation time and other information to ensure that risks are handled in a timely and effective manner.

[0132] S013: Risk monitoring records

[0133] Regularly record information such as changes in risks and treatment effects to ensure that risks are continuously and effectively monitored and managed.

[0134] Specifically, it also includes fraud information identification and interception:

[0135] Through natural language processing and image recognition technology, the system monitors in real time what is displayed on the ATM screen and the conversations that elderly users may be exchanging with fraudsters via their mobile phones. If fraudulent information is identified, such as false transfer prompts, fake bank interfaces, etc., the system immediately intercepts and closes related operations, records the content of the conversation, and sends alerts to the elderly and the bank.

[0136] The steps to identify and block fraudulent information are as follows:

[0137] S014: Information acquisition and preprocessing

[0138] The system collects and monitors all interactive information of bank ATMs in real time, including screen display content, key input, and voice communication of elderly users.

[0139] The collected information is preprocessed to remove noise and irrelevant content and extract key information such as transaction amount, recipient account and transfer notes.

[0140] S015: Feature extraction and model matching

[0141] Machine learning algorithms are used to extract features from preprocessed information and identify characteristic patterns related to fraudulent behavior.

[0142] The extracted features are matched with the pre-trained fraud information recognition model to determine whether there is a fraud risk.

[0143] S016: Real-time risk assessment

[0144] Combine the target user's personal characteristics and current transaction information to conduct real-time risk assessment.

[0145] If the assessment results determine that there is a high risk of fraud, the system will immediately trigger an early warning mechanism.

[0146] S017: Fraud information interception:

[0147] Once an early warning is triggered, the system will automatically intercept transactions suspected of fraud and suspend fund transfer operations. The system will inform elderly users through voice prompts, screen displays, etc. that the transaction has been intercepted and remind users to be vigilant against fraud.

[0148] S018: Subsequent processing and feedback:

[0149] The system will record the intercepted fraud information and related data to provide data support for subsequent risk analysis and model optimization. At the same time, the system will feedback the interception results to users through text messages, phone calls, etc., and guide users to perform subsequent operations, such as re-initiating transactions or reporting alarms.

[0150] S019: Model update and optimization:

[0151] Based on the continuously collected fraud information and user feedback, the fraud information identification model is regularly updated and optimized to improve the identification accuracy and interception efficiency.

[0152] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A business risk management method based on artificial intelligence, characterized by: include Identification and feature extraction of target users, the identification and feature extraction of target users includes: S1: Preliminary judgment of elderly users Obtain the images captured by the camera and perform preliminary identification of the target object in the image through machine learning. If the target object is initially identified as an elderly person, perform specific feature verification; Analyze the captured image and determine the age of the target subject through the recognition module; S2: Feature extraction The age of the subject is determined based on the extracted shooting data containing the picture of the person, and is determined based on the facial features, behavioral features and transaction features; Identify whether the target object is an elderly person. Once identified as an elderly person, the system automatically enters the elderly mode and extracts information such as the user's facial features and behavioral characteristics; It also includes risk behavior identification, which is a risk behavior identification model built based on a deep learning algorithm; It also includes risky transaction identification and early warning, which includes: S3: Real-time transaction monitoring The system captures and records the target user's transaction data in real time, including key information such as withdrawal amount, transaction time, account balance, etc. S4: Transaction Risk Assessment Use machine learning algorithms to conduct risk assessment on transaction data captured in real time; S5: Risk early warning mechanism When the system identifies a high-risk transaction, it will immediately trigger an early warning mechanism, which includes voice prompts, SMS notifications, and notifications to bank staff; S6: Risk treatment and recording Bank staff will intervene in a timely manner and the system will record the processing process and results of each risk warning to provide a basis for subsequent risk analysis and improvements.

2. The business risk processing method based on artificial intelligence according to claim 1 is characterized by: The extracted features mainly include: Facial features: facial recognition technology is used to extract facial features of elderly users, such as wrinkles and skin sagging; Behavioral characteristics: Through cameras and sensors, the behavioral patterns of elderly users when using ATMs, such as operation speed, key sequence, and body posture; Transaction characteristics: Record the transaction history of elderly users, including withdrawal amount, transaction time, account balance and other information.

3. The business risk processing method based on artificial intelligence according to claim 1 is characterized by: The steps for constructing the risk behavior identification model are: S001: The system captures the elderly’s withdrawal behavior data in real time through the optical camera on the ATM; S002: The system uses a pre-trained deep learning model to analyze these behavior data. The model is trained with a large amount of historical data and machine learning algorithms, and can identify abnormal behaviors that do not conform to normal behavior patterns; S003: The system conducts a comprehensive analysis based on the elderly’s historical transaction records, account balances and other information.

4. The business risk processing method based on artificial intelligence according to claim 1 is characterized in that: The transaction risk assessment method for determining real-time transaction monitoring data is as follows: Withdrawal amount: If the withdrawal amount exceeds a certain percentage of the elderly person’s daily transaction habits or account balance, it will be considered a high-risk transaction; Trading time: Withdrawal operations performed during non-regular trading hours will increase the risk of being defrauded and will be considered a high-risk transaction; Transaction frequency: Frequent withdrawals within a short period of time indicate abnormal account activity and will be considered high-risk transactions; Account balance: If the account balance after withdrawal is too low, resulting in the elderly being unable to live a normal life or cope with emergencies, it will be considered a high-risk transaction.

5. The business risk processing method based on artificial intelligence according to claim 1 is characterized by: The early warning steps of the risk early warning mechanism are: S004: Use the voice system of the ATM to send a warning sound to the elderly to remind them to pay attention to transaction safety; S005: Send an early warning message to the elderly person’s registered mobile phone number to remind them to verify the transaction; S006: The system will also push the warning information to the bank's back-end management system for further verification and processing by staff.

6. The business risk processing method based on artificial intelligence according to claim 1 is characterized by: It also includes personalized risk warnings and operation guidance, and the operation steps of the personalized risk warnings and operation guidance are as follows: S007: The system constructs a personalized risk profile based on the target user characteristics, transaction records and risk identification results collected in the previous steps; S008: Generate customized risk alerts for each target user based on the risk profile; S009: Provide personalized operational guidance and continuously optimize the content of risk warnings and operational guidance based on user feedback and real-time risk identification results.

7. The business risk processing method based on artificial intelligence according to claim 1 is characterized by: It also includes risk handling process optimization and recording. The risk handling process optimization includes optimizing the risk handling process and recording the process and results of each risk handling.

8. The business risk processing method based on artificial intelligence according to claim 7 is characterized by: The steps of risk treatment record are: S010: Risk identification records Record risk sources, risk events, potential impacts and other information in detail; S011: Risk Assessment Records Conduct qualitative and quantitative assessment of risks, and record the assessment results, basis and process; S012: Risk management records Record risk management measures, responsible persons, implementation time and other information; S013: Risk monitoring records Regularly record information such as changes in risks and effects of treatment.

Images