Abnormal data behavior detection early warning method and system

Through automated abnormal trading behavior detection methods, the benchmark eigenvector and abnormal evaluation model are used to identify the deviations of transaction volume, frequency and amount, and generate abnormal indexes, which solves the problem that traditional methods are difficult to identify hidden trading behaviors, and achieves efficient and real-time Monitor abnormal transactions to maintain market fairness and transparency.

CN119090501AInactive Publication Date: 2024-12-06临沂市财政事务服务中心
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411089263.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional abnormal trading behavior detection methods rely on manual audits and simple statistical analysis, making it difficult to deal with huge transaction volume and high-frequency data, and it is difficult to identify abnormal trading behaviors with strong concealment, making it difficult to maintain market fairness and transparency.

Method used

By obtaining the asset transaction feature vector, using the predetermined benchmark feature vector for alert judgment, calculating the deviation of transaction volume, frequency and amount, input it into the pre-constructed anomaly evaluation model, generating anomaly index, and comparing it according to the warning threshold. Automatically generate warning information.

Benefits of technology

Improves the accuracy and efficiency of detection, reduces the dependence on manual audits, enables real-time monitoring and identification of obvious and hidden abnormal trading patterns, reduces potential risks, and ensures market fairness and transparency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090501A_ABST
    Figure CN119090501A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of data detection, in particular to an abnormal data behavior detection early warning method and system, which can improve the monitoring efficiency and real-time performance, improve the detection accuracy and maintain the fairness and transparency of the market. The method is used for detecting and identifying abnormal transaction behaviors and performing early warning in an asset transaction process, and comprises the steps of obtaining a target asset transaction feature vector in a set time window; performing warning judgment on the target asset transaction feature vector in a set time window by using a predetermined target asset transaction reference feature vector, and judging whether the target asset transaction feature vector exceeds the target asset transaction reference feature vector or not; in response to the fact that the target asset transaction feature vector exceeds the target asset transaction reference feature vector, obtaining a target asset transaction abnormal feature vector; the target asset transaction anomaly feature vectors comprise the transaction volume, the transaction frequency and the deviation degree between the transaction amount and respective reference values.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data detection, and in particular to a method and system for detecting and warning abnormal data behavior. Background Art

[0002] With the development of financial technology, the speed and scale of asset transactions have increased dramatically. Although this has improved market efficiency, it has also brought new challenges. In particular, the monitoring and early warning of abnormal trading behaviors have become particularly important. Abnormal trading behaviors include insider trading, market manipulation, money laundering and other illegal activities. These behaviors not only undermine the fairness and transparency of the market, but may also cause significant losses to investors and even trigger systemic financial risks.

[0003] Traditional methods for detecting abnormal trading behaviors often rely on manual auditing and simple statistical analysis, which has many shortcomings. First, manual auditing requires a lot of human resources and time costs, and it is difficult to cope with huge trading volumes and high-frequency trading data. Second, simple statistical analysis methods can often only identify obvious abnormal behaviors, but it is difficult to detect abnormal trading behaviors that are more concealed. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides an abnormal data behavior detection and early warning method and system which can improve the efficiency and real-time performance of monitoring, improve the accuracy of detection, and maintain the fairness and transparency of the market.

[0005] In a first aspect, the present invention provides a method for detecting and warning abnormal data behavior, the method being used to detect and identify abnormal transaction behavior and issue a warning during an asset transaction, the method comprising:

[0006] Get the target asset transaction feature vector within the set time window;

[0007] Using a predetermined target asset transaction benchmark feature vector to conduct a warning judgment on the target asset transaction feature vector within a set time window, to judge whether the target asset transaction feature vector exceeds the target asset transaction benchmark feature vector;

[0008] In response to the target asset transaction feature vector exceeding the target asset transaction benchmark feature vector, obtaining the target asset transaction abnormal feature vector; the target asset transaction abnormal feature vector includes the deviation of the transaction volume, transaction frequency and transaction amount from the respective benchmark values;

[0009] Inputting the target asset transaction anomaly feature vector into a pre-built asset transaction anomaly assessment model to obtain the target asset transaction anomaly index;

[0010] The target asset transaction anomaly index is compared with the pre-set warning threshold. If the target asset transaction anomaly index exceeds the warning threshold, an abnormal transaction behavior warning message is generated. Otherwise, no action is taken.

[0011] Furthermore, the method for determining the target asset transaction benchmark feature vector includes:

[0012] Collect historical asset trading behaviors that are the same as the target asset and normal;

[0013] Divide historical asset trading behaviors by time length to obtain multiple historical trading units of the target asset;

[0014] Perform feature extraction on each historical transaction unit to obtain the target asset transaction feature vector of the historical transaction unit;

[0015] Statistical analysis is performed on target asset transaction feature vectors of multiple historical transaction units to obtain target asset transaction benchmark feature vectors.

[0016] Furthermore, factors affecting the setting of the time window include market volatility, transaction frequency, abnormal behavior characteristics, data processing capabilities, response time of the early warning system, and historical data analysis.

[0017] Furthermore, the method for acquiring the target asset transaction abnormality feature vector includes:

[0018] For each transaction feature, the difference between it and the corresponding value of the benchmark feature vector is calculated to obtain the deviation of transaction volume, transaction frequency and transaction amount from their respective benchmark values;

[0019] Set different deviation thresholds to distinguish different degrees of abnormality;

[0020] According to the calculated deviation, the deviation is compared with the set deviation threshold to identify abnormal features;

[0021] The deviations of all abnormal features are integrated into a new vector, which is the target asset transaction abnormal feature vector.

[0022] Furthermore, the method for constructing the asset transaction abnormality assessment model includes:

[0023] Collect historical transaction data, correct errors, missing values ​​and outliers in the data; align and standardize time series data;

[0024] Divide the dataset into training, validation and test sets;

[0025] Selecting a deep learning model as the infrastructure of the asset trading anomaly assessment model, wherein the deep learning model includes logistic regression, random forest, support vector machine, gradient boosting tree and neural network;

[0026] Use the training set to train the selected model, using cross-validation to prevent overfitting; use the validation set to evaluate the model performance;

[0027] Based on the performance of the test set, determine the warning threshold of the anomaly index and balance the false positive rate and the false negative rate;

[0028] Deploy the trained model to the transaction monitoring system.

[0029] Furthermore, the warning threshold is set based on historical data analysis, market trading characteristics, industry regulatory standards, systemic risk prevention and data-driven analysis.

[0030] Furthermore, the target asset transaction feature vector includes transaction volume, transaction frequency and transaction amount.

[0031] On the other hand, the present application also provides an abnormal data behavior detection and early warning system, the system comprising:

[0032] A data acquisition module, used to acquire a target asset transaction feature vector within a set time window, wherein the target asset transaction feature vector includes transaction volume, transaction frequency and transaction amount;

[0033] A warning judgment module uses a predetermined target asset transaction benchmark feature vector to perform a warning judgment on a target asset transaction feature vector within a set time window to determine whether the target asset transaction feature vector exceeds the target asset transaction benchmark feature vector;

[0034] The abnormal feature extraction module obtains the target asset transaction abnormal feature vector in response to the target asset transaction feature vector exceeding the target asset transaction benchmark feature vector; the target asset transaction abnormal feature vector includes the deviation of the transaction volume, transaction frequency and transaction amount from their respective benchmark values;

[0035] The anomaly assessment module inputs the target asset transaction anomaly feature vector into a pre-built asset transaction anomaly assessment model to obtain the target asset transaction anomaly index;

[0036] The warning comparison and generation module compares the target asset transaction anomaly index with the pre-set warning threshold. If the target asset transaction anomaly index exceeds the warning threshold, an abnormal transaction behavior warning message is generated. Otherwise, no action is taken.

[0037] In a third aspect, the present application provides an electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, and the computer program, when executed by the processor, implements the steps of any one of the above methods.

[0038] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps in any one of the above-mentioned methods when executed by a processor.

[0039] Compared with the prior art, the present invention has the following beneficial effects: through automated data processing and analysis, the reliance on manual auditing is significantly reduced, the efficiency of processing huge transaction data is improved, the method can adapt to high-speed and large-scale transaction environments, and solve the limitations of manual auditing in terms of resources and time; the method can complete analysis and judgment within a short time after the transaction occurs, realize real-time monitoring, help to respond quickly, and reduce potential risks;

[0040] By combining the baseline feature vector with the abnormal feature extraction, not only can obvious anomalies be identified, but also more concealed abnormal transaction patterns can be captured through model evaluation, thus improving the accuracy and coverage of detection.

[0041] The preset benchmark feature vectors and warning thresholds can be dynamically adjusted according to market conditions and historical data, ensuring the timeliness and adaptability of monitoring standards and being able to effectively respond to market changes;

[0042] By building an abnormality assessment model, multiple transaction characteristics are comprehensively considered to generate an abnormality index, which can more accurately assess the abnormality of transaction behavior;

[0043] Clear warning threshold settings and automatic warning generation mechanisms ensure that relevant personnel can be notified promptly after abnormal behavior is discovered, providing a valuable time window for subsequent investigation and intervention, helping to prevent potential market risks and protect the interests of investors;

[0044] The method framework has good scalability and can easily integrate more transaction features or adopt more advanced machine learning algorithms to cope with new challenges brought by financial technology innovation and maintain the effectiveness and advancement of the method;

[0045] In summary, this method can improve the efficiency and real-time nature of monitoring, improve the accuracy of detection, and maintain market fairness and transparency through its characteristics of high automation, real-time monitoring, precise detection, and flexible response. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1is a flow chart of the present invention;

[0047] Figure 2 It is a structural diagram of the abnormal data behavior detection and early warning system. DETAILED DESCRIPTION

[0048] In the description of the present application, those skilled in the art should know that the present application can be implemented as a method, an apparatus, an electronic device, and a computer-readable storage medium. Therefore, the present application can be specifically implemented in the following forms: complete hardware, complete software (including firmware, resident software, microcode, etc.), a combination of hardware and software. In addition, in some embodiments, the present application can also be implemented in the form of a computer program product in one or more computer-readable storage media, and the computer-readable storage medium contains computer program code.

[0049] The above-mentioned computer-readable storage medium may adopt any combination of one or more computer-readable storage media. Computer-readable storage media include: electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or devices, or any combination of the above. More specific examples of computer-readable storage media include: portable computer disks, hard disks, random access memories, read-only memories, erasable programmable read-only memories, flash memories, optical fibers, optical disc read-only memories, optical storage devices, magnetic storage devices, or any combination of the above. In the present application, computer-readable storage media can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device, or device.

[0050] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws.

[0051] The present application describes the provided methods, devices, and electronic devices through flowcharts and / or block diagrams.

[0052] It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer-readable program instructions. These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine, and these computer-readable program instructions are executed by a computer or other programmable data processing device to produce a device that implements the functions / operations specified by the boxes in the flowchart and / or block diagram.

[0053] These computer-readable program instructions may also be stored in a computer-readable storage medium that enables a computer or other programmable data processing device to work in a specific manner. In this way, the instructions stored in the computer-readable storage medium produce an instruction device product including functions / operations specified in the blocks in the flowchart and / or block diagram.

[0054] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device, so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby enabling the instructions executed on the computer or other programmable data processing apparatus to provide a process for implementing the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0055] The present application is described below in conjunction with the drawings in the present application.

[0056] Embodiment 1: Figure 1 As shown, the abnormal data behavior detection and early warning method of the present invention specifically includes the following steps:

[0057] S1. Obtain the target asset transaction feature vector within the set time window;

[0058] Step S1 needs to consider the real-time, accuracy and reliability of the data, as well as whether the selected features are sufficient to reflect the important features of the target asset transaction; these feature vectors will provide a basis for subsequent abnormal trading behavior detection, and by comparing with the pre-set benchmark feature vectors, it can be determined whether there is abnormal trading behavior and trigger an early warning;

[0059] The factors affecting the setting of the time window include:

[0060] Market volatility: Different financial markets or asset classes have different natural volatility cycles; therefore, the time window should take into account the market's unique volatility characteristics to adapt to its natural rhythm;

[0061] Trading frequency: The trading activity of an asset directly affects the choice of time window; high-liquidity assets require shorter time windows to capture anomalies in a timely manner, while low-liquidity assets are more suitable for longer time windows;

[0062] Abnormal behavior characteristics: Abnormal trading behaviors occur in a short period of time, requiring a shorter time window for rapid response;

[0063] Data processing capability: In actual operation, the speed of data processing and system resource limitations are also factors to consider; too short a time window will increase the pressure of data processing, while too long a window will reduce the processing burden but will miss immediate abnormal signals;

[0064] Response time of the early warning system: An ideal early warning system issues a warning as soon as possible after abnormal behavior occurs, but responding too quickly will increase the false alarm rate; therefore, the time window needs to balance detection speed and accuracy;

[0065] Historical data analysis: By analyzing historical data, we can identify the time periods when abnormal trading behaviors frequently occur. Setting time windows accordingly can more effectively capture potential abnormal behaviors.

[0066] The target asset transaction feature vector includes:

[0067] Trading volume: the accumulated trading volume of all related transactions within the time window, reflecting the market activity;

[0068] Transaction frequency: Calculates the number of transactions that occurred during the window period, used to assess the intensity of transaction activities;

[0069] Transaction amount: the total amount of all transactions in the summary window, measuring the scale of capital flow;

[0070] In this step, by flexibly setting the time window, we can ensure that we can capture market dynamics in a timely manner, respond quickly to market changes, and improve the timeliness of abnormal behavior identification; the carefully selected transaction feature vectors can fully reflect the core dynamics of asset transactions, provide a solid data foundation for subsequent abnormal analysis, and help accurately identify various abnormal transaction patterns; considering data processing capabilities and system limitations, reasonably setting the time window can not only avoid excessive resource consumption, but also ensure monitoring efficiency, so that the system can effectively respond to the challenges of large amounts of data while ensuring performance; setting the time window in accordance with financial regulatory requirements and industry best practices ensures that the monitoring process is legal and compliant, reduces legal risks, and enhances the reputation and market trust of financial institutions; by comprehensively considering market characteristics, historical data and the response time of the early warning system, the setting of the time window helps to reduce the possibility of false alarms while reducing the false alarm rate, and achieve the best balance between early warning effect and operating costs; the time window setting based on historical data analysis can better adapt to market changes, adjust monitoring strategies in a timely manner, effectively capture potential abnormal trading behaviors, and further maintain market fairness and protect the interests of investors;

[0071] In summary, step S1 is the cornerstone of the abnormal trading behavior detection and early warning mechanism. By scientifically setting the time window and accurately extracting transaction features, it provides accurate and efficient basic data support for subsequent anomaly identification and early warning.

[0072] S2. Using the predetermined target asset transaction benchmark feature vector, a warning judgment is made on the target asset transaction feature vector within the set time window to determine whether the target asset transaction feature vector exceeds the target asset transaction benchmark feature vector;

[0073] The method for constructing the target asset transaction benchmark feature vector includes:

[0074] Collect historical target asset transaction data, covering a sufficiently long time frame and a wide range of market conditions; historical target asset transaction data includes the target asset's transaction volume, transaction frequency, and transaction amount; the time granularity of the data is consistent with the granularity within the set time window to ensure that the benchmark feature vector can accurately reflect the current market status;

[0075] Calculate the baseline feature vector of the target asset's transactions using historical data and calculate the average value of each feature; the average value can provide a basic reference point to measure whether the current transaction is above or below the normal level;

[0076] Calculate the standard deviation of each feature to measure the volatility and variability of the data; standard deviation can help identify abnormal trading behavior that is far from the average level;

[0077] Combine the mean and standard deviation of each feature into a vector, namely the target asset transaction benchmark feature vector;

[0078] The method of using a predetermined target asset transaction benchmark feature vector to conduct a warning judgment on a target asset transaction feature vector within a set time window includes:

[0079] Based on the benchmark feature vector, a warning threshold is set to determine whether the target asset transaction feature vector in the current time window is abnormal; the warning threshold is determined based on the standard deviation and quantile of historical data; the warning threshold is slightly higher than a certain quantile or standard deviation multiple of the benchmark feature vector to ensure that potential abnormal trading behavior can be captured;

[0080] Within the set time window, the target asset transaction feature vector is monitored in real time; whenever a new feature vector is obtained, it is compared with the corresponding benchmark feature vector;

[0081] For each feature, determine whether it exceeds the set warning threshold; if it exceeds the warning threshold, it is considered that the feature vector is abnormal.

[0082] In this step, through automated monitoring and real-time data analysis, the speed and efficiency of abnormal transaction detection are significantly improved, the reliance on manual auditing is reduced, and market changes can be quickly responded to in a high-speed trading environment; the benchmark feature vector accurately calculated using historical data and the warning threshold based on statistics can more accurately identify behaviors that deviate from normal trading patterns; the benchmark feature vector is adjusted dynamically with the market, ensuring that the monitoring standard can reflect the latest market conditions. At the same time, the flexible setting of the warning threshold enhances the model's adaptability to different market volatility; by standardizing transaction features and judging based on quantitative indicators, the objectivity and consistency of the detection process are improved, and the errors caused by subjective judgment are reduced; timely detection and warning of abnormal trading behaviors are helpful to intervene in potential market manipulation, insider trading, money laundering and other illegal activities in advance, protect the interests of investors, maintain market fairness and transparency, and thus effectively reduce systemic financial risks; compared with traditional manual audits, this method greatly saves human resources and time costs, allowing regulators and financial institutions to focus more on high-value analysis and strategy formulation; the implementation of this step not only improves the effectiveness and efficiency of abnormal transaction monitoring, but also provides important technical support for maintaining the healthy and stable development of the financial market.

[0083] S3, in response to the target asset transaction feature vector exceeding the target asset transaction benchmark feature vector, obtaining the target asset transaction abnormal feature vector; the target asset transaction abnormal feature vector includes the deviation of the transaction volume, transaction frequency and transaction amount from their respective benchmark values;

[0084] The method for obtaining the target asset transaction abnormality feature vector includes:

[0085] For each transaction feature, calculate the difference between it and the corresponding value of the benchmark feature vector. Usually, relative deviation or absolute deviation is used to quantify this difference. Calculate the deviation. For transaction volume, calculate the percentage deviation between the actual transaction volume in the current time window and the benchmark transaction volume, which is the transaction volume deviation. For transaction frequency, calculate the difference between the actual transaction frequency in the current time window and the benchmark transaction frequency to obtain the transaction frequency deviation. For transaction amount, calculate the difference between the actual transaction amount in the current time window and the benchmark transaction amount to obtain the transaction amount deviation.

[0086] Set different deviation thresholds to distinguish different degrees of abnormality;

[0087] According to the calculated deviation, the deviation is compared with the set deviation threshold to identify abnormal features;

[0088] Integrate the deviations of all abnormal features into a new vector, which is the abnormal feature vector of target asset transactions;

[0089] In this step, by calculating the deviations of transaction volume, transaction frequency and transaction amount from their respective benchmark values, refined identification of asset transaction anomalies is achieved; it is able to capture slight changes in individual indicators and timely discover potential abnormal behaviors; setting different deviation thresholds allows the system to dynamically adjust the definition of anomalies according to market fluctuations and historical data, thereby improving the flexibility and accuracy of monitoring; integrating the deviations of individual features into an abnormal feature vector enables the system to comprehensively evaluate the degree of abnormality of a transaction or transaction behavior over a period of time from multiple dimensions; compared with manual auditing and simple statistical analysis, automated deviation calculation and threshold comparison greatly improve the speed and efficiency of anomaly detection, and can process a large amount of transaction data in a short time, and timely discover and warn of potential illegal or improper transaction activities; through precise quantitative analysis and reasonable threshold setting, it is possible to find a balance between reducing false positives and false negatives, thereby improving the reliability and practicality of the monitoring system; the implementation of step S3 not only enhances the scientificity and accuracy of abnormal transaction monitoring, but also effectively improves the ability to protect the healthy operation of the financial market, reduces systemic risks, and safeguards the interests of investors and market fairness.

[0090] S4. Inputting the target asset transaction abnormality feature vector into a pre-built asset transaction abnormality assessment model to obtain the target asset transaction abnormality index;

[0091] The model calculates based on the input abnormal feature vector and outputs the target asset transaction abnormality index; the target asset transaction abnormality index is the model's quantitative assessment of the degree of abnormality of transaction behavior. The abnormality index output by the model is used for early warning judgment in subsequent steps;

[0092] The method for constructing the asset transaction abnormality assessment model includes:

[0093] Collect a large amount of historical transaction data from financial data sources, correct errors, missing values ​​and outliers in the data; align and standardize time series data to ensure data quality;

[0094] The dataset is divided into training set, validation set and test set with a ratio of 70%, 15% and 15% to ensure the effectiveness of model training and the objectivity of evaluation;

[0095] Selecting a deep learning model as the infrastructure of the asset trading anomaly assessment model, wherein the deep learning model includes logistic regression, random forest, support vector machine, gradient boosting tree and neural network;

[0096] Use the training set to train the selected model, using cross-validation to prevent overfitting;

[0097] Use the validation set to evaluate model performance, focusing on indicators including accuracy, recall, F1 score, and AUC-ROC curve;

[0098] Based on the performance of the test set, determine the warning threshold of the anomaly index, balance the false alarm rate and the missed alarm rate, and ensure the practicality of the early warning system;

[0099] Deploy the trained model to the transaction monitoring system.

[0100] In this step, the abnormal feature vectors of target asset transactions are automatically processed through deep learning models, which significantly improves the speed and accuracy of abnormal transaction detection, reduces the reliance on manual review, and effectively copes with the challenges of large data volumes and high-frequency transactions. The model can comprehensively consider the deviation of multi-dimensional features such as transaction volume, frequency, and amount, and quantitatively evaluate the abnormal degree of transaction behavior, providing a more accurate abnormal index, which helps to capture a variety of hidden abnormal behaviors including insider trading and market manipulation. Based on a strict model construction process, including data preprocessing, model selection and training, performance evaluation, and threshold determination, the high sensitivity and low false alarm rate of the early warning system are ensured, the false alarm rate and missed alarm rate are balanced, and the practicality and reliability of the early warning are improved. The continuous monitoring and maintenance mechanism of the model supports regular updates or optimizations based on market dynamics and new data feedback, maintains the model's ability to identify emerging transaction patterns and abnormal behaviors, and enhances the adaptability and foresight of the system. The application of this model can effectively monitor and warn of illegal trading activities, protect the interests of investors, maintain market order, enhance the fairness and transparency of the financial market, help prevent systemic financial risks, and promote healthy and stable market development.

[0101] S5. Compare the target asset transaction abnormality index with the pre-set warning threshold. If the target asset transaction abnormality index exceeds the warning threshold, generate abnormal transaction behavior warning information. Otherwise, no action is taken.

[0102] The warning threshold is a key criterion for determining whether a transaction behavior is abnormal. The basis for setting the warning threshold includes:

[0103] Historical data analysis: By analyzing historical abnormal transaction cases, typical abnormal transaction behavior characteristics and the distribution of abnormal indexes can be determined, which helps to set the basis for early warning thresholds;

[0104] Market transaction characteristics: The frequency, amount and volatility of transactions in different markets can vary significantly, and the warning threshold should take these differences into account;

[0105] Industry regulatory standards: Financial regulators will issue regulatory guidance on abnormal transactions, including the definition of abnormal transactions, typical characteristics, and standards and requirements for early warning processing; the setting of early warning thresholds needs to comply with these regulatory standards to ensure that the system is consistent with regulatory requirements;

[0106] Systemic risk prevention: The setting of early warning thresholds needs to take into account the impact of abnormal transactions on the overall market stability and systemic risks;

[0107] Data-driven analysis: Using machine learning models and statistical methods, the warning thresholds are dynamically adjusted and optimized based on real-time and historical data; the models can help identify changing patterns and new abnormal behavior characteristics, and then adjust the warning thresholds to improve the accuracy and efficiency of detection.

[0108] In this step, by setting thresholds based on comprehensive historical data analysis, market characteristics, regulatory requirements and data-driven methods, the real abnormal trading behavior can be identified more accurately, false positives and false negatives can be reduced, and the accuracy and reliability of the monitoring system can be improved. The data-driven dynamic adjustment mechanism enables the warning threshold to adapt to market changes and reflect the latest trading behavior patterns in a timely manner, ensuring that a rapid response can be made when abnormal trading first appears, effectively shortening the time difference from the occurrence of abnormalities to their discovery. The threshold setting that incorporates industry regulatory standards ensures the legality of the operation of the monitoring system, helps financial institutions and regulatory agencies to effectively comply with relevant laws and regulations, and reduce compliance risks. The threshold setting that considers systemic risk prevention helps to identify and warn of abnormal trading activities that may pose a threat to market stability in advance, protect market fairness and transparency, safeguard investor interests, and avoid or mitigate systemic financial risks. The automated and accurate warning system reduces reliance on manual audits, saves a lot of human resources and time costs, and enables financial institutions to allocate resources more efficiently and focus on handling abnormal events that really need attention. Reasonable warning threshold setting not only strengthens the monitoring efficiency of abnormal trading behaviors, but also plays a vital role in ensuring the healthy operation of the market, promoting financial stability, improving compliance efficiency, and optimizing resource utilization.

[0109] Embodiment 2: Figure 2 As shown, the abnormal data behavior detection and early warning system of the present invention specifically includes the following modules:

[0110] A data acquisition module, used to acquire a target asset transaction feature vector within a set time window, wherein the target asset transaction feature vector includes transaction volume, transaction frequency and transaction amount;

[0111] A warning judgment module uses a predetermined target asset transaction benchmark feature vector to perform a warning judgment on a target asset transaction feature vector within a set time window to determine whether the target asset transaction feature vector exceeds the target asset transaction benchmark feature vector;

[0112] The abnormal feature extraction module obtains the target asset transaction abnormal feature vector in response to the target asset transaction feature vector exceeding the target asset transaction benchmark feature vector; the target asset transaction abnormal feature vector includes the deviation of the transaction volume, transaction frequency and transaction amount from their respective benchmark values;

[0113] The anomaly assessment module inputs the target asset transaction anomaly feature vector into a pre-built asset transaction anomaly assessment model to obtain the target asset transaction anomaly index;

[0114] The warning comparison and generation module compares the target asset transaction anomaly index with the pre-set warning threshold. If the target asset transaction anomaly index exceeds the warning threshold, an abnormal transaction behavior warning message is generated. Otherwise, no action is taken.

[0115] The system greatly reduces the reliance on manual audits through automated data acquisition and processing processes, and can quickly process large-scale and high-frequency transaction data, solving the problems of high labor costs and long response times.

[0116] The system can complete data analysis within the real-time or near-real-time time window of transactions, detect abnormal trading behaviors in a timely manner, and help to take quick measures to prevent the spread of risks;

[0117] Through the abnormal feature extraction module and advanced anomaly assessment model, it is possible to identify more than just obvious anomalies, improving the accuracy and coverage of monitoring, and helping to uncover more complex market manipulation and illegal trading strategies;

[0118] The preset baseline feature vectors and warning thresholds can be continuously adjusted and optimized according to market dynamics and historical data to ensure that the system can adapt to market changes and maintain the accuracy and effectiveness of warnings;

[0119] The modular design makes the system easy to maintain and upgrade. Each module can be adjusted or have new functions added as needed to cope with new types of abnormal trading behaviors that may appear in the future. By effectively monitoring and timely warning abnormal transactions, the system can help maintain a fair competition environment in the market, protect the interests of investors, reduce market imbalances caused by improper trading behaviors, and thus improve the stability and transparency of the entire financial market.

[0120] In summary, the system can improve the efficiency and real-time nature of monitoring, increase the accuracy of detection, and maintain market fairness and transparency through its features of high automation, real-time monitoring, precise detection, and flexible response.

[0121] The various variations and specific embodiments of the abnormal data behavior detection and early warning method in the aforementioned embodiment 1 are also applicable to the abnormal data behavior detection and early warning system of the present embodiment. Through the aforementioned detailed description of the abnormal data behavior detection and early warning method, those skilled in the art can clearly know the implementation method of the abnormal data behavior detection and early warning system of the present embodiment, so for the sake of brevity of the specification, it will not be described in detail here.

[0122] In addition, the present application also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are respectively connected via a bus. When the computer program is executed by the processor, each process of the above-mentioned method for controlling output data is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0123] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for detecting and warning abnormal data behavior, characterized in that: The method is used to detect and identify abnormal transaction behavior and issue an early warning during an asset transaction, and the method includes: Get the target asset transaction feature vector within the set time window; Using a predetermined target asset transaction benchmark feature vector to conduct a warning judgment on the target asset transaction feature vector within a set time window, to judge whether the target asset transaction feature vector exceeds the target asset transaction benchmark feature vector; In response to the target asset transaction feature vector exceeding the target asset transaction benchmark feature vector, obtaining the target asset transaction abnormal feature vector; the target asset transaction abnormal feature vector includes the deviation of the transaction volume, transaction frequency and transaction amount from the respective benchmark values; Inputting the target asset transaction anomaly feature vector into a pre-built asset transaction anomaly assessment model to obtain the target asset transaction anomaly index; The target asset transaction anomaly index is compared with the pre-set warning threshold. If the target asset transaction anomaly index exceeds the warning threshold, an abnormal transaction behavior warning message is generated. Otherwise, no action is taken.

2. The abnormal data behavior detection and early warning method according to claim 1, characterized in that: The method for determining the target asset transaction benchmark feature vector includes: Collect historical asset trading behaviors that are the same as the target asset and are normal; Divide historical asset trading behaviors by time length to obtain multiple historical trading units of the target asset; Perform feature extraction for each historical transaction unit to obtain the target asset transaction feature vector of the historical transaction unit; Statistical analysis is performed on target asset transaction feature vectors of multiple historical transaction units to obtain target asset transaction benchmark feature vectors.

3. The abnormal data behavior detection and early warning method according to claim 1, characterized in that: The factors affecting the setting of the time window include market volatility, transaction frequency, abnormal behavior characteristics, data processing capabilities, response time of the early warning system and historical data analysis.

4. The abnormal data behavior detection and early warning method according to claim 1, characterized in that: The method for obtaining the target asset transaction abnormality feature vector includes: For each transaction feature, the difference between it and the corresponding value of the benchmark feature vector is calculated to obtain the deviation of transaction volume, transaction frequency and transaction amount from their respective benchmark values; Set different deviation thresholds to distinguish different degrees of abnormality; According to the calculated deviation, the deviation is compared with the set deviation threshold to identify abnormal features; The deviations of all abnormal features are integrated into a new vector, which is the target asset transaction abnormal feature vector.

5. The abnormal data behavior detection and early warning method according to claim 1, characterized in that: The method for constructing the asset transaction abnormality assessment model includes: Collect historical transaction data, correct errors, missing values ​​and outliers in the data; align and standardize time series data; Divide the dataset into training, validation and test sets; Selecting a deep learning model as the infrastructure of the asset trading anomaly assessment model, wherein the deep learning model includes logistic regression, random forest, support vector machine, gradient boosting tree and neural network; Use the training set to train the selected model, using cross-validation to prevent overfitting; use the validation set to evaluate the model performance; Based on the performance of the test set, determine the warning threshold of the anomaly index and balance the false positive rate and the false negative rate; Deploy the trained model to the transaction monitoring system.

6. The abnormal data behavior detection and early warning method according to claim 1, characterized in that: The basis for setting the warning threshold includes historical data analysis, market trading characteristics, industry regulatory standards, systemic risk prevention and data-driven analysis.

7. The abnormal data behavior detection and early warning method according to claim 1, characterized in that: The target asset transaction feature vector includes transaction volume, transaction frequency and transaction amount.

8. An abnormal data behavior detection and early warning system, characterized in that: The system comprises: A data acquisition module, used to acquire a target asset transaction feature vector within a set time window, wherein the target asset transaction feature vector includes transaction volume, transaction frequency and transaction amount; A warning judgment module uses a predetermined target asset transaction benchmark feature vector to perform a warning judgment on a target asset transaction feature vector within a set time window to determine whether the target asset transaction feature vector exceeds the target asset transaction benchmark feature vector; The abnormal feature extraction module obtains the target asset transaction abnormal feature vector in response to the target asset transaction feature vector exceeding the target asset transaction benchmark feature vector; the target asset transaction abnormal feature vector includes the deviation of the transaction volume, transaction frequency and transaction amount from their respective benchmark values; The anomaly assessment module inputs the target asset transaction anomaly feature vector into a pre-built asset transaction anomaly assessment model to obtain the target asset transaction anomaly index; The warning comparison and generation module compares the target asset transaction anomaly index with the pre-set warning threshold. If the target asset transaction anomaly index exceeds the warning threshold, an abnormal transaction behavior warning message is generated. Otherwise, no action is taken.

9. An abnormal data behavior detection and early warning electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, characterized in that: When the computer program is executed by the processor, the steps in the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 7 are implemented.

Citation Information

Cited By

  • Abnormal transaction identification and early warning system based on financial time sequence characteristics

    CN120031572A

  • Abnormal transaction behavior detection method and system and readable storage medium

    CN121304174A