Consumption financial data whole-process comprehensive processing method and system

Abstract

The invention provides a consumption financial data whole-process comprehensive processing method and system, and relates to the technical field of consumption financial data processing. Comprising the steps of evaluating whether current consumption financial data of a target user triggers risk control rules in a risk control rule pool; updating a risk index pool containing a plurality of reference user categories and corresponding risk indexes by using current consumption financial data of a target user triggering a risk control rule; performing risk identification on the current consumption financial data of the target user which does not trigger the risk control rule by adopting a machine learning method to obtain a first risk score; obtaining a second risk score based on the matching condition of the current consumption financial data of the target user which does not trigger the risk control rule and the reference user category in the updated risk index pool; and obtaining a comprehensive risk score of the target user based on the first risk score and the second risk score, and performing risk monitoring on the current consumption financial data of the target user. The accuracy of risk assessment is improved.

Description

Technical Field

[0001] This invention belongs to the field of consumer finance data processing technology, and in particular relates to a comprehensive method and system for processing consumer finance data throughout the entire process. Background Technology

[0002] As an important branch of the financial industry, consumer finance faces diverse and complex risks due to its inherent business characteristics. Consumer finance encompasses various risk types, including credit risk, market risk, and operational risk. These risks permeate every stage of loan disbursement, fund recovery, and customer service. Therefore, establishing a comprehensive monitoring and early warning mechanism is crucial for consumer finance.

[0003] Before providing consumer finance services to a user, it is necessary to assess the risks involved. Currently, this assessment often requires manual processing, which introduces a degree of subjectivity. If the assessment is biased and incorrectly evaluates the user's risk profile, it will increase the credit risk for the financial institution and could even cause serious economic losses.

[0004] With the development of artificial intelligence, risk assessment of users before providing consumer financial services has become possible using AI technology. However, when using AI technology to assess the risk of target users, if a target user is determined to be of high risk, their data is directly discarded, without making full use of the discarded data.

[0005] Furthermore, there is currently a lack of a comprehensive method for processing user consumer finance data throughout the entire process, which causes many inconveniences for technical personnel when performing data repair, anomaly detection, access control, and leak prevention. Summary of the Invention

[0006] To overcome the shortcomings of the prior art, this invention provides a comprehensive processing method and system for the entire process of consumer finance data. When it is determined that a target user has triggered risk control rules and thus poses a significant risk, the high-risk target user data is not discarded directly. Instead, it is used in the update of the risk index pool. The risk index corresponding to the benchmark user category that matches the high-risk target user in the risk index pool is increased, and then applied to the risk assessment of low-risk target users in the later stage, thereby improving the accuracy of risk assessment.

[0007] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: The first aspect of this invention provides a comprehensive method for processing consumer finance data throughout the entire process.

[0008] A comprehensive method for processing consumer finance data throughout the entire process includes the following steps: Obtain the target user's current consumer finance data and perform preprocessing; Assess whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool; By utilizing the current consumer finance data of target users who trigger risk control rules, the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, is updated. Machine learning methods are used to identify risks in the current consumer finance data of target users who have not triggered risk control rules, and a first risk score is obtained. The second risk score is obtained by matching the current consumer finance data of target users who have not triggered risk control rules with the benchmark user category in the updated risk index pool. Based on the first risk score and the second risk score, a comprehensive risk score for the target user is obtained, and risk monitoring is performed on the target user's current consumer finance data.

[0009] The second aspect of this invention provides a comprehensive method and system for processing consumer finance data throughout the entire process.

[0010] A comprehensive data processing system for the entire consumer finance process includes: The data acquisition module is configured to: acquire the target user's current consumer finance data and perform preprocessing; The trigger module is configured to: evaluate whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool; The risk control strategy update module is configured to update the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, using the current consumer finance data of the target user that triggered the risk control rules. The first risk score calculation module is configured to: use machine learning methods to identify risks in the current consumer finance data of target users who have not triggered risk control rules, and obtain the first risk score; The second risk score calculation module is configured to: obtain the second risk score based on the matching of the target user's current consumer finance data that has not triggered risk control rules with the benchmark user category in the updated risk index pool; The comprehensive risk score calculation module is configured to: obtain the comprehensive risk score of the target user based on the first risk score and the second risk score, and monitor the risk of the target user's current consumer finance data. A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the comprehensive processing method for the entire consumer finance data process as described in the first aspect of the present invention.

[0011] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the comprehensive processing method for the entire process of consumer finance data as described in the first aspect of the present invention.

[0012] The above one or more technical solutions have the following beneficial effects: This invention provides a comprehensive method and system for processing consumer finance data throughout the entire process. When it is determined that a target user has triggered a risk control rule and thus poses a significant risk, the high-risk target user data is not discarded directly. Instead, it is applied to the risk control strategy, increasing the risk index corresponding to the benchmark user category that matches the high-risk target user in the risk control strategy. This index is then applied to the subsequent risk assessment of low-risk target users, thereby improving the accuracy of risk assessment.

[0013] This invention employs machine learning methods to identify the risk of current consumer finance data of target users who have not triggered risk control rules, obtaining a first risk score. Based on the matching between the current consumer finance data of target users who have not triggered risk control rules and the benchmark user category in the updated risk control strategy, a second risk score is obtained. Based on the first and second risk scores, a comprehensive risk score for the target user is obtained. This method fully utilizes the data of high-risk target users identified based on risk control rules, so that the impact of the identification results is reflected in the risk index corresponding to the benchmark user category in the risk control strategy, thereby achieving a more accurate risk assessment for low-risk target users.

[0014] This invention also includes data repair, anomaly detection, access control, and leakage prevention processing for the current consumer finance data of target users who have not triggered risk control rules. The proposed comprehensive processing method for the entire consumer finance data process is more convenient for technical personnel to operate.

[0015] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0016] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0017] Figure 1 This is a flowchart of the method in Example 1.

[0018] Figure 2 This is a diagram illustrating the comprehensive data processing architecture for consumer finance in Example 1. Detailed Implementation

[0019] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0021] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0022] Example 1 This embodiment discloses a comprehensive method for processing consumer finance data throughout the entire process. This method is mainly applied in the consumer finance field, effectively improving the intelligence and compliance of risk control management, and reducing the occurrence of data leakage and risk events.

[0023] like Figure 1 As shown, a comprehensive method for processing consumer finance data throughout the entire process includes the following steps: Obtain the target user's current consumer finance data and perform preprocessing; Assess whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool; By utilizing the current consumer finance data of target users who trigger risk control rules, the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, is updated. Machine learning methods are used to identify risks in the current consumer finance data of target users who have not triggered risk control rules, and a first risk score is obtained. The second risk score is obtained by matching the current consumer finance data of target users who have not triggered risk control rules with the benchmark user category in the updated risk index pool. Based on the first risk score and the second risk score, a comprehensive risk score for the target user is obtained, and risk monitoring is performed on the target user's current consumer finance data.

[0024] Next, we will combine the appendix Figure 2 The method described in this embodiment will be explained in detail.

[0025] (I) Data Acquisition and Preprocessing Obtain the target user's current consumer finance data and perform preprocessing, specifically including: The target user's current consumer finance data includes all current moment data from user registration to application, from application to credit granting, from credit granting to borrowing, and from borrowing to repayment. The preprocessing includes structuring the data, understanding the semantics of fields and mapping labels, and cleaning and standardizing the data.

[0026] More specifically, the steps include: A: Accessing multi-source external data Function: The goal of this step is to receive data from external data sources (such as API interfaces, databases, file uploads, etc.) and import it into the system. External data may be structured (such as databases), semi-structured (such as JSON, XML), or unstructured (such as text, images).

[0027] Implementation method: API Access: Communicate with external systems via API to obtain real-time data. Access to the data interface requires HTTP requests (such as GET and POST).

[0028] Database access: Establish connections with external databases (such as MySQL and PostgreSQL), execute SQL queries, and retrieve data.

[0029] File upload: Supports file uploads from FTP, SFTP, and Web, and supports formats such as CSV, JSON, and Excel. It reads file content using a file parsing tool and formats it into data that can be processed in memory.

[0030] Data format conversion: Ensure that all external data can be converted into a uniform structured format, such as tables or JSON objects, for subsequent processing.

[0031] Interaction: The output of this step is a formatted data stream, which enters the B-format recognition and adaptation template construction, laying the foundation for subsequent processing.

[0032] B: Format recognition and template adaptation Function: This step ensures that all external data meets the expected format requirements and builds an adaptation template so that data from different sources can be uniformly entered into the system.

[0033] Implementation method: Format recognition: Algorithms (such as regular expressions, type inference, etc.) are used to detect different data formats to ensure that all incoming data conforms to a certain format.

[0034] Adaptive template construction: Design corresponding parsing templates for different data sources (such as APIs, databases, and files). For example, when parsing a JSON file, it is necessary to define the field structure, and when parsing a CSV file, it is necessary to determine the column names and delimiters.

[0035] Field mapping: Maps fields in the input data to standard field names to ensure that the data can be processed uniformly.

[0036] Interaction: After the data is processed uniformly through the adapted template, it enters the C field semantic parsing and tag mapping, which facilitates subsequent semantic parsing and tag processing.

[0037] C: Field semantic parsing and label mapping Function: Perform semantic analysis on fields in the data and assign standardized labels to each field for subsequent business processing.

[0038] Implementation method: Field semantic parsing: Using NLP techniques or rule-based matching methods, the meaning of field names is parsed and converted into standardized names. For example, the "Name" field is parsed as "name", and "Date of Birth" is parsed as "birth_date".

[0039] Tag mapping: Add tags to each field to conform to the standardized business model. Tags may include field type (e.g., text, date), field purpose (e.g., user information, transaction information), etc.

[0040] Interaction: Data after field semantic parsing flows into D for data cleaning and standardization to ensure that the data is free of dirty data and meets standardization requirements.

[0041] D: Data cleaning and standardization Function: The purpose of this step is to improve data quality, remove invalid and non-standard data, and ensure that all data conforms to the predetermined format and quality standards.

[0042] Implementation method: Deduplication: Identify and delete duplicate records to avoid data redundancy.

[0043] Missing value handling: Missing values ​​in the data can be handled by interpolation (such as mean imputation) or by deleting rows containing missing values.

[0044] Outlier detection: Detecting outliers in the data using statistical methods (such as Z-score, IQR, etc.) and deciding whether to delete or replace these outliers.

[0045] Formatting: Use a consistent date format (e.g., YYYY-MM-DD) and number format (e.g., currency precision) to ensure consistency.

[0046] Standardization: Standardizing numerical data to make data of different dimensions comparable (e.g., using Z-score standardization).

[0047] Interaction: After the cleaned and standardized data enters E, does it trigger risk control rules? Check if risk control is required.

[0048] (ii) Triggering judgment Assess whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool, specifically including: Retrieve the risk control rules in the risk control rule pool at the current moment; The target user's current consumer finance data is compared with the risk control rules to determine whether there are any anomalies in the target user's current consumer finance data; When the target user's current consumer finance data is abnormal, it is determined that a risk control rule in the risk control rule pool has been triggered.

[0049] Further: The specific construction process of the risk control rules in the risk control rule pool at the current moment is as follows: Based on user behavior pattern rules, abnormal consumption rules, borrowing frequency rules, credit score fluctuation rules, and geographical location rules, the initial risk control rules in the pre-set risk control rule pool are used. Collect current consumer finance data of all target users who have triggered the initial risk control rules at the current moment, extract common features, and identify abnormal patterns; The obtained common features and abnormal patterns are organized into a new feature set; Based on the new feature set, the initial risk control rules are updated to form new rules and the new rules are verified to obtain the risk control rules in the risk control rule pool at the current moment; The abnormal patterns include borrowing beyond the preset frequency, borrowing beyond the preset amount, abnormal consumption behavior, abnormal geographical location changes, and credit score inquiries beyond the preset frequency.

[0050] Further: Based on the new feature set, the initial risk control rules are updated to form new rules, which are then validated to obtain the risk control rules in the risk control rule pool at the current moment, specifically including: Features that exist in the new feature set but not in the initial risk control rules are used as new rules; By backtesting with historical data, the effectiveness of the new rules is verified, and the applicability of the new rules to historical data and their ability to identify risky users are analyzed. Newly validated rules will be incorporated into the risk control rule pool. Based on the performance of the new rules and feedback from practical applications, the weight of the new rules in the risk control rule pool is dynamically adjusted to update the initial risk control rules.

[0051] Further: Based on user behavior pattern rules, abnormal consumption rules, borrowing frequency rules, credit score fluctuation rules, and geographical location rules, initial risk control rules are pre-set in the risk control rule pool, including: Behavioral pattern rules: These include classifying changes to contact information or address that exceed a set frequency as high-risk behavior.

[0052] Abnormal spending rules: These rules define a user's spending behavior as abnormal if it exceeds a set range compared to the historical average.

[0053] Loan frequency rules: This includes setting a user's loan frequency as abnormal if they apply for multiple loans exceeding a set limit within a set time period.

[0054] Credit score fluctuation rules: If a user's credit score fluctuates beyond a set range within a set time period, it is considered an abnormal credit score.

[0055] Geographic location rules: If a user's consumption or borrowing location changes or becomes abnormally frequent within a set time period, it is considered an abnormal geographic location.

[0056] Specifically: 1. Generate initial risk control rules, specifically: - Behavioral pattern rules: If a user frequently changes their contact information or address within a short period of time, it is set as a high-risk behavior.

[0057] - Abnormal spending rules: If a user's spending behavior suddenly exceeds the historical average level, it is set as abnormal spending.

[0058] - Loan frequency rule: If a user applies for multiple high-amount loans within a month, it will be set as an abnormal loan frequency.

[0059] - Credit Score Fluctuation Rules: If a user's credit score fluctuates significantly in a short period of time, it is set as an abnormal credit score.

[0060] - Geographic location rules: If a user's consumption or borrowing location changes frequently or abnormally, it will be set as an abnormal geographic location.

[0061] 2. Analyze the characteristic patterns that trigger risk control rules: First, collect the current consumer finance data of all target users who have triggered risk control rules, and then conduct in-depth analysis of this data.

[0062] - By using machine learning and data mining techniques, common features and abnormal patterns in these data can be extracted, such as frequent high-interest loans, abnormal consumption behavior, and frequent credit score inquiries.

[0063] 3. Construct a feature set for abnormal behavior: - Organize these common features and abnormal patterns into a feature set to construct new risk control rules.

[0064] - For example, frequent high-amount borrowing, unusual changes in geographical location, and multiple loan applications within a short period of time can be set as new risk control rules.

[0065] 4. Verify and optimize the new rules: - Verify the effectiveness of the new rules by backtesting with historical data, and analyze the applicability of the new rules to historical data and their ability to identify risky users.

[0066] - Based on the backtesting results, further optimize the rules to ensure their accuracy and effectiveness.

[0067] 5. Dynamically adjust rule weights: - Based on the performance of the new rules and feedback from practical applications, dynamically adjust the weight of the new rules in the risk control rule pool.

[0068] - For example, if a new rule excels at identifying high-risk users, its priority and weight in the rule pool can be increased.

[0069] 6. Iterative updates of rules: - Regularly evaluate and update risk control rules to ensure their timeliness and accuracy.

[0070] - Continuously optimize and iterate risk control rules based on the latest user behavior data and changes in the risk index pool.

[0071] Figure 2 In the middle, E: Has the risk control rule been triggered? Function: Determines whether the data meets the preset risk control rules and decides whether to carry out further risk control processing.

[0072] Implementation method: Rule engine: Based on preset risk control rules, a rule engine (such as Drools) assesses whether the data meets risk standards. Rules may be based on transaction amount, user behavior patterns, etc.

[0073] Historical data comparison: Compare current data with historical data (such as user behavior, credit scores, etc.) to determine if there are any anomalies.

[0074] Risk control model assistance: Using machine learning models (such as decision trees, random forests, etc.) to predict data and determine whether there is any risk.

[0075] Interaction: Yes: If the data triggers the risk control rules, the data enters the F risk control mark + gray list + real-time alarm.

[0076] No: If the data does not trigger the risk control rules, the data flows to G to optimize SQL construction and write to the main table, and continues subsequent processing.

[0077] F: Risk control tagging + gray warehouse + real-time alarms Function: To mark, store, and provide real-time alerts for data that triggers risk control rules.

[0078] Implementation method: Risk control labeling: Data is labeled as "high risk" through a rules engine or manual intervention.

[0079] Gray database storage: High-risk data is stored in a gray database, awaiting subsequent manual intervention or automated processing. The gray database is an isolated database and will not affect the operation of the main database.

[0080] Real-time alerts: Send real-time alerts via message queues (such as Kafka) and monitoring tools (such as Prometheus) to notify relevant personnel to handle risks.

[0081] Interaction: Data marked with risk control is fed into K1 intelligent risk control rule generation and dynamic optimization to improve the accuracy and adaptability of risk control strategies.

[0082] K1: Intelligent Risk Control Rule Generation and Dynamic Optimization Function: Based on historical data and real-time feedback, automatically generate and optimize risk control rules to improve the system's ability to respond to risks.

[0083] Implementation method: Automated rule generation: Automatically discover potential risk patterns and generate new risk control rules through machine learning algorithms (such as decision trees, cluster analysis, etc.).

[0084] Dynamic rule optimization: Based on real-time data and risk control effectiveness, existing rules are adjusted. The rule engine dynamically adjusts the weight and application strategy of each rule.

[0085] Interaction: New or optimized rules are fed into K2. A / B testing provides feedback on the effectiveness of the rules, which are then verified through A / B testing.

[0086] K2: Rule Advantages and Disadvantages A / B Test Feedback Function: Compare the effects of different risk control rules and select the best rule set.

[0087] Implementation method: A / B testing: Divide the data into multiple groups, apply different sets of rules to each group, and evaluate their performance.

[0088] Effectiveness evaluation: The effectiveness of the rules is evaluated using metrics such as false positive rate, false negative rate, and interception rate.

[0089] Feedback: Test results are automatically fed back to the rules engine for optimization and adjustment of rules.

[0090] Interaction: After A / B testing, the rules enter K3 for automatic rule optimization and global optimization, and then undergo final optimization.

[0091] K3: Automatic rule optimization and global optimization Function: Optimize risk control rules based on A / B test results to ensure the best risk control effect.

[0092] Implementation method: Rule optimization algorithms: These use machine learning methods (such as genetic algorithms, gradient descent, etc.) to automatically optimize existing rules.

[0093] Global optimization: By integrating different rule sources (such as historical data, real-time data, etc.), the parameters of the rules are adjusted globally to improve the overall performance.

[0094] Interaction: The optimized rules are entered into the H data database and then analyzed further through I data lineage tracing and data visualization.

[0095] (III) Data entry Perform the following steps G and H on the current consumer finance data of target users who have not triggered risk control rules.

[0096] G optimizes SQL construction and writing to the main table.

[0097] H: Data import complete Function: Writes cleaned, labeled, and optimized data into the main database, providing a foundation for subsequent data analysis.

[0098] Implementation method: Database optimization: Use distributed database technology to ensure efficient and scalable data writing.

[0099] Data transaction management: Ensure the integrity of data when it is written and use transaction mechanisms (such as ACID) to prevent data loss.

[0100] Interaction: After the data is entered into the database, the data is analyzed and visualized.

[0101] (iv) Secondary use of high-risk user data + risk assessment of low-risk users By utilizing the current consumer finance data of target users who trigger risk control rules, the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, is updated. Specifically, this includes: Based on the current consumer finance data of the target users who triggered the risk control rules, machine learning methods were used to determine the profile of the target users who triggered the risk control rules. By comparing the similarity between the target user profile that triggers the risk control rule and multiple benchmark user categories, the benchmark user category with the highest similarity to the target user profile that triggers the risk control rule is found. The risk index corresponding to the benchmark user category with the highest similarity to the target user profile that triggers the risk control rule will be increased, and the risk control strategy will be updated.

[0102] The target user profile that triggers the risk control rules is obtained by using machine learning methods to extract features from the current consumer finance data of the target users who trigger the risk control rules and then classifying the target users who trigger the risk control rules. The baseline user category and its corresponding risk index are obtained through pre-setting; The risk index corresponding to the benchmark user category that has the highest similarity to the target user profile that triggers the risk control rule will be increased. The specific increase method can be to increase the corresponding original risk index by one level or to increase a certain risk value according to the degree of risk.

[0103] The technical implementation of the above process includes three aspects: similarity matching, risk index adjustment, and dynamic risk assessment. - Similarity matching: Machine learning models are used to perform similarity matching between the target user profile that triggers risk control rules and the benchmark user category to find the benchmark user category with the highest similarity.

[0104] - Risk index adjustment: For the benchmark user category with the highest similarity, increase its risk index, for example, by raising the original risk index by one level or by adding a certain risk value according to the degree of risk.

[0105] - Dynamic risk assessment: Based on the latest data and model feedback, we continuously monitor and dynamically adjust the risk index of benchmark user categories to ensure the accuracy and timeliness of risk assessment.

[0106] Furthermore, based on the matching of the current consumer finance data of target users who have not triggered risk control rules with the benchmark user categories in the updated risk index pool, a second risk score is obtained, which specifically includes: Based on the current consumer finance data of target users who have not triggered risk control rules, machine learning methods are used to determine the profile of target users who have not triggered risk control rules. By comparing the similarity between the target user profile that has not triggered risk control rules and multiple benchmark user categories, the benchmark user category with the highest similarity to the target user profile that has not triggered risk control rules is found. The risk index corresponding to the benchmark user category with the highest similarity to the target user profile that has not triggered risk control rules will be used as the second risk score.

[0107] The target user profile for those who have not triggered risk control rules is obtained by using machine learning methods to extract features from the current consumer finance data of these users and then classifying them. - Feature extraction: Using machine learning algorithms (such as cluster analysis, decision trees, etc.) to extract features from the current consumer finance data of target users who have not triggered risk control rules, and extract the main features and behavioral patterns.

[0108] - User classification: Cluster and classify the extracted features to divide users into different risk levels or categories.

[0109] - Risk Assessment: By comparing the risk profiles of users who have not triggered risk control rules with the risk index of benchmark user categories, we assess their risk level and ensure that potential risks can be accurately identified even when risk control rules have not been triggered.

[0110] Furthermore, based on the first risk score and the second risk score, a comprehensive risk score for the target user is obtained, specifically including: The first risk score and the second risk score are weighted and combined to obtain the target user's comprehensive risk score.

[0111] O1: Deep Learning and Data Mining Function: Uses deep learning technology to intelligently mine data and identify potential patterns and trends.

[0112] Implementation method: Deep learning: using neural networks (such as convolutional neural networks (CNNs), recurrent neural networks (RNNs) to deeply mine big data and identify potential patterns in the data.

[0113] Data mining: Applying data mining algorithms (such as clustering, classification, and association rules) to analyze data and extract valuable information.

[0114] Interaction: The results of deep learning and data mining are passed to O2 pattern recognition and data intelligence mining to further analyze the behavior of the data.

[0115] O2: Pattern Recognition and Intelligent Data Mining Function: Through pattern recognition technology, further analyze the data, identify behavioral patterns, and provide intelligent decision support.

[0116] Implementation method: Pattern recognition: Using pattern recognition algorithms (such as support vector machines (SVM) and k-means clustering) to identify trends and patterns in data.

[0117] Intelligent mining: Based on the results of pattern recognition, it provides intelligent mining tools to support automated decision-making.

[0118] Interaction: The pattern recognition and mining results are passed to O3 for abnormal behavior identification and risk control strategy generation, generating new risk control strategies.

[0119] O3: Abnormal Behavior Detection and Risk Control Strategy Generation Function: Automatically generate targeted risk control strategies by identifying abnormal behavior.

[0120] Implementation method: Abnormal behavior identification: By analyzing user or system behavior, identify whether there are abnormal behavior patterns. For example, user login time, transaction frequency, transaction amount, etc.

[0121] Strategy Generation: Based on the results of abnormal behavior identification, new risk control strategies are automatically generated and applied to protect the system from potential threats.

[0122] Interaction: New risk control strategies are integrated into P1 for cross-system intelligent linkage, ensuring cross-system risk control collaboration.

[0123] (v) Joint processing P1: Cross-system intelligent linkage Function: Enables intelligent linkage between multiple systems, ensuring that risk control strategies are executed synchronously across multiple platforms and systems.

[0124] Implementation method: Intelligent linkage: Real-time data sharing and risk control strategy synchronization between different systems are achieved through API and message queue technology.

[0125] System integration: Ensure seamless data exchange between various systems and maintain consistency during policy execution.

[0126] Interaction: After linkage, the data flows into P2 for multi-system data sharing and interface integration, further improving the efficiency of data collaboration.

[0127] P2: Multi-system data sharing and interface integration Function: Supports data sharing and interface integration between multiple systems, ensuring efficient data flow across different systems.

[0128] Implementation method: Data sharing: Data from multiple systems can be shared and integrated through interface technologies such as RESTful API and GraphQL.

[0129] Interface integration: Ensure that the interface standards of different systems are consistent to facilitate data exchange.

[0130] Interaction: After interface integration, the data flows to P3 for cross-platform risk control collaboration and decision-making, enabling cross-platform risk control and decision execution.

[0131] P3: Cross-platform risk control collaboration and decision-making Function: Enables collaborative risk control decisions across multiple platforms to ensure overall risk control and strategy execution.

[0132] Implementation method: Cross-platform collaboration: By sharing data and risk control strategies across multiple platforms, consistency and efficiency in the execution of risk control strategies can be achieved across various platforms.

[0133] Decision execution: Based on cross-platform risk control strategies, make overall decisions and execute them to ensure that risks are controllable.

[0134] Interaction: The collaborative results across platforms flow to Q1 for quantitative decision-making and game theory optimization, further optimizing the decision-making model.

[0135] (vi) Strategy Selection Q1: Quantitative Decision Making and Game Theory Optimization Function: Apply quantitative decision-making and game theory optimization methods to select the optimal risk control strategy.

[0136] Implementation method: Quantitative decision-making: Using mathematical models and statistical methods to quantify decisions and evaluate the effects and risks of different decisions.

[0137] Game theory optimization: Optimize the selection of multiple risk control strategies through game theory models to find the optimal strategy combination.

[0138] Interaction: The results of quantitative decision-making and game theory optimization enter Q2. Decision quantification and global optimization achieve comprehensive optimization of decision-making.

[0139] Q2: Decision Quantification and Global Optimization Function: To optimize the decision-making process globally and ensure the optimality of each decision in the global environment.

[0140] Implementation method: Global optimization: Apply global optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) to comprehensively optimize risk control decisions.

[0141] Decision evaluation: Evaluate different decision options to ensure the selection of the optimal strategy.

[0142] Interaction: The optimized decision-making scheme enters Q3 for optimal strategy selection and multi-dimensional evaluation, further selecting the best strategy.

[0143] Q3: Optimal Strategy Selection and Multidimensional Evaluation Function: Select the optimal risk control strategy based on multi-dimensional evaluation.

[0144] Implementation method: Multidimensional assessment: Evaluate the strategy across multiple dimensions (such as economic benefits, operational complexity, compliance, etc.).

[0145] Optimal strategy selection: Based on the evaluation results, select the risk control strategy that best meets the current business needs.

[0146] Interaction: The optimal strategy enters R1 for real-time feedback and intelligent tuning, providing real-time feedback and optimization of the strategy.

[0147] (vii) Strategy Optimization R1: Real-time feedback and intelligent optimization Function: Through a real-time feedback mechanism, continuously adjust risk control strategies to cope with dynamically changing risk environments.

[0148] Implementation method: Real-time feedback: The monitoring system obtains the execution effect of risk control strategies in real time and adjusts the strategies based on the results.

[0149] Intelligent tuning: Using machine learning algorithms (such as reinforcement learning) to self-adjust the policy in order to improve its performance.

[0150] Interaction: Real-time feedback results are fed into R2 for real-time feedback analysis and automatic tuning, allowing for further optimization.

[0151] R2: Real-time feedback analysis and automatic tuning Function: Analyze real-time feedback results and automatically optimize risk control strategies.

[0152] Implementation method: Feedback Analysis: Analyze real-time feedback data to evaluate the effectiveness of the current risk control strategy.

[0153] Automatic optimization: Based on feedback analysis results, the risk control strategy is automatically adjusted to improve its response capabilities.

[0154] Interaction: The optimized results enter the R3 self-learning mechanism and rule optimization for self-learning and further optimization.

[0155] R3: Self-learning mechanism and rule optimization Function: Through a self-learning mechanism, further optimize rules and risk control strategies.

[0156] Implementation method: Self-learning mechanism: Utilizing machine learning models, risk control rules are automatically optimized by learning from historical data.

[0157] Rule optimization: Based on the self-learning results, existing rules are optimized and adjusted to better adapt to emerging risk patterns.

[0158] Interaction: The optimization results of the self-learning mechanism are incorporated into the entire risk control system for final strategy adjustment and application.

[0159] (viii) Data repair processing I: Data lineage tracing and data visualization Function: Tracks the flow and changes of data, and visualizes the data to help decision-makers understand it.

[0160] Implementation method: Data lineage tracing: Records the entire process of data from access to storage, helping to trace the source and history of data changes.

[0161] Data visualization: Displaying data through charts, dashboards, etc., allows users to intuitively understand the data situation.

[0162] Interaction: The visualized results are further processed and suggested by the J Data Repair Suggestion Engine.

[0163] J: Data Repair Suggestion Engine Function: Provides repair suggestions based on data issues (such as anomalies, missing values, etc.) to optimize data quality.

[0164] Implementation method: Repair suggestions: Based on data quality checks (such as missing values ​​and outliers), repair suggestions are generated, which may include filling missing values ​​and correcting outliers.

[0165] Rule Engine: Automatically provides repair suggestions based on preset rules and machine learning algorithms.

[0166] Interaction: Repair suggestions are provided to users through a J1 question-and-answer rule-based interface, offering better repair solutions.

[0167] J1: Question-and-answer rule building interface Function: Allows users to input repair rules through an interactive interface and build custom data repair processes.

[0168] Implementation method: Interface design: Allow users to select repair rules and strategies through a simple graphical interface.

[0169] Rule building: Automatically generate repair rules based on user input and apply them to the data.

[0170] (ix) Data anomaly detection and handling L1: Anomaly Detection and Fault Tolerance Switching Module Function: Used to detect anomalies in the data and perform fault tolerance processing when anomalies occur to ensure system stability.

[0171] Implementation method: Anomaly detection: Detecting anomalies in data using predefined rules or machine learning models. Anomalies may include field mutations, data inconsistencies, logical errors, etc.

[0172] Fault tolerance mechanism: When an anomaly is detected, the system can automatically switch to a backup template or recovery strategy to avoid data loss or error transmission.

[0173] Interaction: When L1 detects an anomaly, it will enter L2 to check if the field has mutated or the value is abnormal, and further determine whether processing is required.

[0174] L2: Is there a field mutation or an abnormal value? Function: Checks whether a field in the data has undergone a sudden change or whether the field value is abnormal. If so, the system performs the appropriate processing.

[0175] Implementation method: Field mutation detection: Compare current data with historical data to detect any mutations in field values. For example, the user's name field may mutate, or the values ​​of other sensitive fields may be abnormal.

[0176] Value anomaly detection: By setting thresholds (such as range checks), this detects whether values ​​in the data fall within the expected range. For example, negative transaction amounts or extreme values.

[0177] Interaction: Yes: If a field mutation or value abnormality is detected, the data will enter L3 to switch to an alternative mapping template and log records for processing.

[0178] No: If there are no anomalies, the data will continue to remain in the main process.

[0179] L3: Switching to alternative mapping template + logging Function: In case of anomalies, switch to the backup mapping template and log the changes to ensure that data can still be processed normally when anomalies occur.

[0180] Implementation method: Alternate mapping template: When data anomalies are detected, the system will switch to an alternative template or processing logic. For example, alternative field mappings can be used, or data cleaning rules can be adjusted.

[0181] Log recording: Records the entire process of exception handling to facilitate subsequent auditing and problem tracing.

[0182] Interaction: After handling the exception, the system will return the data to the main process and continue to execute subsequent tasks.

[0183] (x) Access Control Processing M1: Data Access Auditing and Access Control Function: Ensures that data access complies with access control regulations, performs access auditing, and protects sensitive data.

[0184] Implementation method: Access auditing: Recording and tracking data access behavior to ensure all data operations are traceable. For example, recording who accessed which data and when.

[0185] Access control: Control access permissions to data based on different user roles to ensure that only authorized users can access sensitive data.

[0186] Interaction: Access control and auditing data flow to M2 field-level authorization and access chain evidence storage, further refining access control.

[0187] M2: Field-level authorization and access chain evidence storage Function: Implement access control for each data field to ensure that the data access chain is traceable.

[0188] Implementation method: Field-level authorization: Set different access permissions for each data field. For example, only management can access a user's detailed personal information, while general employees can only access basic information.

[0189] Access chain evidence: Records the access chain for each field to ensure that access to and manipulation of data can be traced back to the specific user and time.

[0190] Interaction: This information is passed to the M3 Compliance Audit and Risk Identification module to ensure that data access complies with compliance requirements.

[0191] M3: Compliance Audit and Risk Identification Module Function: Ensures all data access operations comply with requirements and identifies potential risks.

[0192] Implementation method: Compliance audit: Based on records of data access and operations, ensure that all operations comply with national or regional laws and regulations (such as GDPR, CCPA, etc.).

[0193] Risk identification: Identify potential compliance risks and data breach risks through the results of compliance audits.

[0194] Interaction: The results of compliance audits and risk identification provide a reference for subsequent decisions, ensuring that compliance is not violated during the N1 federal data sharing and encrypted access phase.

[0195] (xi) Leakage prevention measures N1: Federal Data Sharing and Encrypted Access Functionality: Supports data sharing and encrypted access without exposing raw data, ensuring data privacy and security.

[0196] Implementation method: Federated learning: Federated learning technology allows multiple systems to share a model without exchanging raw data, ensuring data privacy.

[0197] Encrypted access: Using encryption technology to protect the security of data during the sharing process and prevent data leakage or unauthorized access.

[0198] Interaction: The results of federal data sharing are incorporated into the N2 data watermarking and leakage tracing mechanism to ensure data traceability.

[0199] N2: Data Watermarking and Leakage Tracing Mechanism Function: Embed watermarks in data to ensure that data breaches can be traced back to their source.

[0200] Implementation method: Data watermarking: Embedding specific identifiers or watermarks into data to identify the source and change records of the data.

[0201] Leakage tracing: When data is leaked, the source of the leak can be traced through watermarks, allowing for accountability.

[0202] Interaction: Watermarking technology ensures that traceability information is not lost during data circulation. The data flows to O1 deep learning and data mining for further analysis.

[0203] Core innovations: 1. AI-driven intelligent perception and identification mechanism for data anomalies Innovation: This method uses AI algorithms (deep learning, machine learning rule engines) to perceive and identify abnormal behavior in data in real time, thus proactively identifying potential problems in the early stages of the data flow. This innovative data processing approach breaks through the limitations of traditional static detection methods, enabling it to dynamically adapt to data changes and adjust processing strategies in real time.

[0204] Technical advantages: Compared with traditional data anomaly detection methods, AI-driven perception mechanisms have higher accuracy and flexibility, and can continuously improve the recognition rate through self-learning and optimization.

[0205] 2. Multidimensional data processing and fault tolerance mechanism based on rule engine Innovation: This method employs an AI rule engine to handle anomalous data and automatically switches between multiple fault-tolerance strategies based on different data types or severity. This rule engine does not rely solely on preset rules but adaptively optimizes processing strategies based on data changes and contextual information.

[0206] Technical advantages: Compared with traditional fixed rule processing methods, AI-driven rule engines can intelligently select the optimal strategy based on various factors such as historical data and trend changes, thereby improving fault tolerance and adaptability.

[0207] 3. Anomaly repair and self-learning optimization closed loop Innovation: This method proposes a closed-loop system for anomaly repair and self-learning optimization. After identifying anomalous data, the system not only reacts immediately but also continuously adjusts and optimizes the anomaly identification algorithm based on feedback during processing. This closed-loop mechanism ensures that the system can continuously improve data processing quality in practical applications.

[0208] Technical advantages: This innovation breaks the limitation of traditional systems that can only "passively handle" problems, and has the ability to self-repair and self-optimize, enabling it to cope with complex and dynamically changing data environments.

[0209] 4. Multi-layered data governance and risk early warning system Innovation: This method introduces a multi-layered risk control and early warning mechanism into the data governance process. Through AI models, the system can perform quality control, anomaly identification, risk assessment, and provide risk warnings at multiple levels. This multi-layered governance approach can more accurately grasp potential problems in the data flow and avoid missing risks at a single level.

[0210] Technical advantages: Compared with traditional single data processing and risk assessment methods, a multi-layered risk warning system can better adapt to complex business needs and data environments, thereby providing more comprehensive risk prevention and control.

[0211] 5. Adaptive Data Stream Processing and Decision Support Innovation: This method can automatically adjust the data stream processing flow based on the characteristics and changes of the input data. This adaptive processing capability enables the system to make efficient decisions in different data scenarios and quickly switch to an appropriate processing mode when anomalies occur. This approach significantly improves the system's intelligence level and processing efficiency.

[0212] Technical advantages: Traditional systems usually require predefined strict rules, while adaptive processing can flexibly adjust based on the actual situation of the data, improving the intelligence of the decision support system, and is particularly suitable for large-scale and diverse data processing scenarios.

[0213] 6. Full lifecycle data governance and anomaly recovery mechanism Innovation: This method not only covers anomaly identification during the data access phase, but also provides governance and anomaly recovery support throughout the entire data flow lifecycle. From data access to processing, storage, and data feedback, each stage can be effectively monitored and optimized. This full lifecycle governance ensures continuous improvement in data quality.

[0214] Technical advantages: This full lifecycle processing approach differs from traditional single-stage governance. It ensures that data maintains high quality from start to finish, reducing subsequent problems caused by neglecting anomalies in a single stage.

[0215] 7. Intelligent decision support system with real-time feedback and automatic adjustment Innovation: This method features a real-time feedback mechanism that continuously receives feedback during system operation and automatically adjusts decisions accordingly. This innovation enables the system to dynamically respond to changes in the external environment and optimize the data processing decision-making process in real time.

[0216] Technical advantages: Compared with traditional static decision support systems, this system can be flexibly adjusted and self-corrected through real-time feedback, making decision-making more forward-looking and flexible, and adapting to rapidly changing data environments.

[0217] Advantages and effects: This embodiment achieves adaptive risk control decisions through techniques such as deep learning, machine learning, and game theory optimization, and has the following advantages: 1. Precise identification and prevention of data anomalies Advantages: The system uses AI algorithms (such as deep learning and machine learning) to intelligently perceive and identify data anomalies in real time, accurately capturing abnormal behaviors in data streams. This method is more flexible and accurate than traditional rule-based detection methods, and can identify latent anomalies that traditional methods overlook.

[0218] Results: Anomaly detection is enabled during the data input stage, preventing the further escalation of potential problems and significantly improving the accuracy and efficiency of data processing. It reduces the need for manual intervention, making the data processing process more automated and intelligent.

[0219] 2. Highly adaptive fault tolerance and optimization capabilities Advantages: The system can dynamically adjust its fault tolerance strategy based on data changes, selecting the most suitable strategy for the current data processing scenario through adaptive algorithms. Compared to traditional static rules, the AI-driven fault tolerance mechanism can flexibly cope with complex and ever-changing data environments.

[0220] Results: When abnormal data occurs, the system can quickly switch to the optimal fault-tolerance scheme, ensuring that data streams are repaired promptly when problems arise. Through this dynamic optimization, the system can guarantee efficient data stream processing, improving overall data quality and system reliability.

[0221] 3. Anomaly repair and self-learning closed-loop mechanism Advantages: This method introduces a self-learning optimization closed-loop mechanism. After identifying and repairing abnormal data, the system will optimize itself based on real-time feedback. The algorithm continuously improves its identification and repair capabilities by accumulating feedback data during each processing step, thereby enhancing the overall processing performance.

[0222] Results: The system can gradually improve its recognition accuracy and repair efficiency over long-term application, eventually forming a "self-evolving" intelligent processing system. This avoids the rigidity and lag of fixed rules in traditional methods and adapts to the rapidly changing data environment.

[0223] 4. Multi-layered risk prevention and intelligent early warning Advantages: The system is designed with a multi-layered risk prevention and control system. AI algorithms monitor each link in the data flow in real time and provide intelligent early warnings when anomalies occur. This multi-dimensional risk monitoring can promptly identify potential risk points and prevent the spread of problems.

[0224] Results: Through layered filtering and multi-dimensional evaluation, the system can accurately identify and warn of data risks, reducing the incidence of risk events. In complex data flow scenarios, this multi-layered prevention and control mechanism effectively ensures the stability and security of data processing.

[0225] 5. Data quality assurance throughout the entire lifecycle Advantages: The data governance proposed in this method is not limited to anomaly identification in a single stage, but rather spans the entire data lifecycle, ensuring quality and optimizing each step from data input, processing, storage to feedback. AI technology is used to adjust the governance strategy in real time, ensuring that each step meets the highest data processing standards.

[0226] Results: This full lifecycle governance ensures high data quality throughout its entire lifecycle, reducing data quality issues caused by oversights in single stages. It is particularly effective when handling large amounts of dynamically changing data and can adapt to complex application scenarios.

[0227] 6. Efficient decision support and automatic adjustment mechanism Advantages: The system provides real-time decision support through AI algorithms, automatically adjusting processing strategies based on changes in the data stream and offering data-driven decision recommendations. Compared to traditional manual decision-making or fixed rules, AI-driven decision support can react quickly based on actual data conditions.

[0228] Results: This automatically adjusting decision-making mechanism reduces human error and delays, enabling appropriate responses to complex data situations in a short time. The immediacy and accuracy of decisions significantly improve the efficiency of data stream processing and the system's response speed.

[0229] 7. Enhance the system's intelligence level Advantages: By combining various AI technologies such as deep learning, machine learning, and rule engines, this system can continuously optimize its processing flow and improve its intelligence level. The system's capabilities in automatic identification, decision-making, and optimization far surpass traditional manual methods.

[0230] Results: The system reduces human intervention during data processing and optimizes processing results through continuous learning and feedback. With prolonged use, the system's intelligence level continuously improves, ultimately forming an autonomous, efficient, and stable intelligent data processing platform.

[0231] 8. Reduce operating costs and reliance on human resources Advantages: Traditional data anomaly identification and processing rely on manual rule configuration and intervention, requiring significant human resources and time. This method, however, leverages AI-driven characteristics to reduce reliance on manual intervention and achieves large-scale automated processing.

[0232] Results: By automating identification, repair, and decision support, it reduces manual intervention and operational costs, lowering the company's reliance on human resources. It effectively improves work efficiency while reducing error rates and processing delays caused by human error.

[0233] 9. Strong scalability and adaptability Advantages: The system design boasts high scalability and adaptability, capable of accommodating various data stream types and application scenarios. Whether it's changes in data volume, data type, or application environment, the system can be flexibly adjusted and optimized.

[0234] Results: Whether dealing with big data scenarios or diverse data sources, the system can seamlessly integrate and provide efficient processing capabilities. As application scenarios expand, the system can continuously adapt to new demands and consistently provide efficient services.

[0235] Example 2 This embodiment discloses a comprehensive method and system for processing consumer finance data throughout the entire process.

[0236] A comprehensive data processing system for the entire consumer finance process includes: The data acquisition module is configured to: acquire the target user's current consumer finance data and perform preprocessing; The trigger module is configured to: evaluate whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool; The rule pool update module is configured to: mark the current consumer finance data of the target user that triggers the risk control rule to obtain risk control mark data; generate new risk control rules based on the risk control mark data; determine whether to add the new risk control rules to the risk control rule pool based on the effect of the new risk control rules; and update the risk control rule pool. The risk control strategy update module is configured to update the risk control strategy, which includes multiple benchmark user categories and corresponding risk indices, using the current consumer finance data of the target user that triggered the risk control rule. The first risk score calculation module is configured to: use machine learning methods to identify risks in the current consumer finance data of target users who have not triggered risk control rules, and obtain the first risk score; The second risk score calculation module is configured to: obtain the second risk score based on the matching of the current consumer finance data of the target user who has not triggered the risk control rules with the benchmark user category in the updated risk control strategy; The comprehensive risk score calculation module is configured to: obtain the comprehensive risk score of the target user based on the first risk score and the second risk score, and monitor the risk of the target user's current consumer finance data. Example 3 The purpose of this embodiment is to provide a computer-readable storage medium.

[0237] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the comprehensive processing method for the entire consumer finance data process as described in Embodiment 1 of this disclosure.

[0238] Example 4 The purpose of this embodiment is to provide an electronic device.

[0239] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the comprehensive processing method for the entire consumer finance data process as described in Embodiment 1 of this disclosure.

[0240] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0241] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0242] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A comprehensive method for processing consumer finance data throughout the entire process, characterized in that, Includes the following steps: Obtain the target user's current consumer finance data and perform preprocessing; Assess whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool; By utilizing the current consumer finance data of target users who trigger risk control rules, the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, is updated. Machine learning methods are used to identify risks in the current consumer finance data of target users who have not triggered risk control rules, and a first risk score is obtained. The second risk score is obtained by matching the current consumer finance data of target users who have not triggered risk control rules with the benchmark user category in the updated risk index pool. Based on the first risk score and the second risk score, a comprehensive risk score for the target user is obtained, and risk monitoring is performed on the target user's current consumer finance data.

2. The comprehensive processing method for the entire consumer finance data process as described in claim 1, characterized in that, Obtain the target user's current consumer finance data and perform preprocessing, specifically including: The target user's current consumer finance data includes all current moment data from user registration to application, from application to credit granting, from credit granting to borrowing, and from borrowing to repayment. The preprocessing includes structuring the data, understanding the semantics of fields and mapping labels, and cleaning and standardizing the data.

3. The comprehensive processing method for the entire consumer finance data process as described in claim 1, characterized in that, Assess whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool, specifically including: Retrieve the risk control rules in the risk control rule pool at the current moment; The target user's current consumer finance data is compared with the risk control rules to determine whether there are any anomalies in the target user's current consumer finance data; When the target user's current consumer finance data is abnormal, it is determined that a risk control rule in the risk control rule pool has been triggered. or, The specific construction process of the risk control rules in the risk control rule pool at the current moment is as follows: Based on user behavior pattern rules, abnormal consumption rules, borrowing frequency rules, credit score fluctuation rules, and geographical location rules, the initial risk control rules in the pre-set risk control rule pool are used. Collect current consumer finance data of all target users who triggered the initial risk control rules before the current moment, extract common features, and identify abnormal patterns; The obtained common features and abnormal patterns are organized into a new feature set; Based on the new feature set, the initial risk control rules are updated to form new rules and the new rules are verified to obtain the risk control rules in the risk control rule pool at the current moment; The abnormal patterns include borrowing beyond the preset frequency, borrowing beyond the preset amount, abnormal consumption behavior, abnormal geographical location changes, and credit score inquiries beyond the preset frequency. or, Based on the new feature set, the initial risk control rules are updated to form new rules, which are then validated to obtain the risk control rules in the risk control rule pool at the current moment, specifically including: Features that exist in the new feature set but not in the initial risk control rules are used as new rules; By backtesting with historical data, the effectiveness of the new rules is verified, and the applicability of the new rules to historical data and their ability to identify risky users are analyzed. Newly validated rules will be incorporated into the risk control rule pool. Based on the performance of the new rules and feedback from practical applications, the weight of the new rules in the risk control rule pool is dynamically adjusted to update the initial risk control rules. or, Based on user behavior pattern rules, abnormal consumption rules, borrowing frequency rules, credit score fluctuation rules, and geographical location rules, initial risk control rules are pre-set in the risk control rule pool, including: Behavioral pattern rules: These include classifying users who change their contact information or address more frequently than the set frequency as high-risk behavior. Abnormal spending rules: These include classifying a user's spending behavior as abnormal if it exceeds a set range compared to the historical average. Loan frequency rules: This includes setting a user's loan frequency as abnormal if they apply for multiple loans exceeding a set limit within a set time period. Credit score fluctuation rules: This includes defining a user's credit score as abnormal if it fluctuates beyond a set range within a set time period. Geographic location rules: If a user's consumption or borrowing location changes or becomes abnormally frequent within a set time period, it is considered an abnormal geographic location.

4. The comprehensive processing method for the entire consumer finance data process as described in claim 1, characterized in that, By utilizing the current consumer finance data of target users who trigger risk control rules, the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, is updated. Specifically, this includes: Based on the current consumer finance data of the target users who triggered the risk control rules, machine learning methods were used to determine the profile of the target users who triggered the risk control rules. By comparing the similarity between the target user profile that triggers the risk control rule and multiple benchmark user categories, the benchmark user category with the highest similarity to the target user profile that triggers the risk control rule is found. The risk index corresponding to the benchmark user category with the highest similarity to the target user profile that triggers the risk control rule will be increased, and the risk control strategy will be updated.

5. The comprehensive processing method for the entire consumer finance data process as described in claim 1, characterized in that, The second risk score is obtained by matching the current consumer finance data of target users who have not triggered risk control rules with the baseline user category in the updated risk index pool. Specifically, this includes: Based on the current consumer finance data of target users who have not triggered risk control rules, machine learning methods are used to determine the profile of target users who have not triggered risk control rules. By comparing the similarity between the target user profile that has not triggered risk control rules and multiple benchmark user categories, the benchmark user category with the highest similarity to the target user profile that has not triggered risk control rules is found. The risk index corresponding to the benchmark user category with the highest similarity to the target user profile that has not triggered risk control rules will be used as the second risk score.

6. The comprehensive processing method for the entire consumer finance data process as described in claim 1, characterized in that, Based on the first risk score and the second risk score, a comprehensive risk score for the target user is obtained, which includes: The first risk score and the second risk score are weighted and combined to obtain the target user's comprehensive risk score.

7. The comprehensive processing method for the entire consumer finance data process as described in claim 1, characterized in that, It also includes data repair, anomaly detection, access control, and data leakage prevention for the current consumer finance data of target users who have not triggered risk control rules.

8. A comprehensive data processing method and system for the entire consumer finance process, characterized in that: include: The data acquisition module is configured to: acquire the target user's current consumer finance data and perform preprocessing; The trigger module is configured to: evaluate whether the target user's current consumer finance data triggers risk control rules in the risk control rule pool; The risk control strategy update module is configured to update the risk index pool, which includes multiple benchmark user categories and corresponding risk indices, using the current consumer finance data of the target user that triggered the risk control rules. The first risk score calculation module is configured to: use machine learning methods to identify risks in the current consumer finance data of target users who have not triggered risk control rules, and obtain the first risk score; The second risk score calculation module is configured to: obtain the second risk score based on the matching of the target user's current consumer finance data that has not triggered risk control rules with the benchmark user category in the updated risk index pool; The comprehensive risk score calculation module is configured to: obtain the comprehensive risk score of the target user based on the first risk score and the second risk score, and monitor the risk of the target user's current consumer finance data.

9. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the comprehensive processing method for the entire consumer finance data process as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the comprehensive processing method for the entire process of consumer finance data as described in any one of claims 1-7.

Images