Transaction risk control method and system

Abstract

The invention relates to the technical field of transaction risk control, in particular to a transaction risk control method and system, and the method comprises the following steps: obtaining platform transaction information through a transaction risk control system; setting a transaction data transmission and arrangement mechanism to arrange and detect the platform transaction information, and obtaining a transaction information transmission result; establishing an information feature analysis model according to the platform transaction information, and analyzing a transaction information transmission result by using the information feature analysis model to obtain transaction feature information of an online transaction platform; and analyzing a platform transaction condition based on the transaction feature information and the platform transaction information so as to guarantee risk control and safe operation of an online transaction platform. According to the transaction risk control method and system, the mathematical model is utilized, the transaction data transmission, arrangement, analysis and application processes are optimized, the user experience and the platform operation efficiency can be improved, and the safety and stability of the platform can be effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of transaction risk control, and in particular to a transaction risk control method and system. Background Art

[0002] With the advancement and application of cutting-edge mobile Internet technologies such as big data, cloud computing, artificial intelligence and blockchain, the integration boundary of science and technology and finance has been continuously broadened, stimulating the innovation and transformation of transaction management models. Faced with the increasingly complex and diversified challenges of the trading market environment, the traditional transaction risk management and early warning mechanism has shown passivity and lag, which has brought challenges to the trading platform in terms of operation strategy, technological innovation and risk prevention and control. The existing transaction data risk monitoring methods mainly rely on the business risk supervision system, which has problems such as low degree of data processing automation and incomplete coverage. At the same time, there is a lack of transaction data identification models, detection technologies and adjustment mechanisms, which cannot achieve real-time acquisition, efficient processing, intelligent analysis and automatic early warning of online transaction information.

[0003] With the development of online trading platforms, it is necessary to systematically integrate massive data resources inside and outside the enterprise and integrate big data and Internet technology into traditional trading management methods. The purpose is to build a trading risk management method with data as the core driving force, strengthen the real-time monitoring and intelligent early warning capabilities of platform trading risks, and build a comprehensive and multi-level risk prevention and response mechanism, thereby improving the accuracy of online trading risk assessment, the timeliness of early warning, and the intelligent level of risk disposal. Summary of the invention

[0004] In view of the shortcomings of existing methods and the needs of practical applications, in order to achieve comprehensive and automated processing of transaction data, a collation, detection and analysis recognition model is added to the transaction data prevention and control method to achieve real-time acquisition, efficient processing, intelligent analysis and automatic warning of online transaction information, thereby improving the risk assessment accuracy, warning timeliness and intelligence level of the online transaction platform. On the one hand, the present invention provides a transaction risk control method, which includes the following steps: obtaining platform transaction information through a transaction risk control system; setting a transaction data transmission and collation mechanism to collate and detect the platform transaction information, and obtaining the transaction information transmission result; establishing an information feature analysis model based on the platform transaction information, and using the information feature analysis model to analyze the transaction information transmission result to obtain the transaction feature information of the online transaction platform; analyzing the platform transaction status based on the transaction feature information and the platform transaction information to ensure the risk control and safe operation of the online transaction platform. The present invention comprehensively manages and analyzes the platform transaction information based on the transaction risk control method and system, improves the risk control and safe operation level of the online transaction platform, and provides strong guarantees for the steady development of the transaction platform and user safety transactions.

[0005] Optionally, the setting of the transaction data transmission and sorting mechanism to sort and detect the platform transaction information and obtain the transaction information transmission result includes: setting the transaction data transmission and sorting mechanism according to the platform transaction information, wherein the setting of the transaction data transmission and sorting mechanism includes a transmission node transaction information model, a transaction data loss degree analysis model, and a transaction information integration model. The present invention introduces the transmission node transaction information model, the transaction data loss degree analysis model, and the transaction information integration model, which can significantly improve the performance and effect of the transaction data transmission and sorting mechanism, and provide solid support for the effective operation of the transaction risk control method.

[0006] Optionally, the transaction data transmission and sorting mechanism is set to sort and detect the platform transaction information, including: using the transmission node transaction information model to obtain the propagation matrix of any node in the transaction information transmission process; using the transaction data loss degree analysis model to obtain the loss of platform transaction information; constructing a transaction information integration model based on the transmission node transaction information model and the transaction data loss degree analysis model, and using the transaction information integration model to obtain the transmission information set of the online transaction platform. The present invention sets up a transaction data transmission and sorting mechanism, which can significantly improve the actual performance of the transaction risk control method, and is conducive to ensuring the steady development and safe transactions of the online transaction platform.

[0007] Optionally, the transport node transaction information model satisfies the following relationship:

[0008]

[0009] Among them, H i represents the propagation matrix of each node in the process of transaction information transmission, softmax represents the Softmax function, h i represents the matrix of the i-th node, Y i represents the adjacency matrix of the ith node, B i represents the degree matrix of the i-th node matrix, S i (t) represents the input feature matrix corresponding to the timestamp of the i-th node, Q i (t) represents the weight matrix of the i-th node, P i (t) represents the offset vector corresponding to the timestamp of the i-th node. The transport node transaction information model of the present invention comprehensively considers multiple key factors and dynamic changes, and provides an accurate, flexible and efficient information dissemination model for the trading platform, which helps to improve the operational efficiency and risk management level of the trading platform.

[0010] Optionally, the transaction data loss degree analysis model satisfies the following relationship:

[0011]

[0012] Among them, LOSS represents the loss coefficient of transaction information, N represents the total number of nodes of transaction information, η represents the similarity coefficient between the propagation matrix of the i-th node and the sample propagation matrix, l represents the distance between the propagation matrix of the i-th node and the sample propagation matrix, and α represents the propagation matrix distance threshold. The present invention quantitatively evaluates the loss of transaction information during transmission through the loss coefficient, provides an intuitive and comparable evaluation standard for the trading platform, and helps to more accurately understand the loss of transaction data and the overall impact.

[0013] Optionally, the transaction information integration model satisfies the following relationship:

[0014]

[0015] Among them, H represents the transmission matrix of transaction information, LOSS represents the loss coefficient of transaction information, N represents the total number of nodes of transaction information, and H i The invention represents the propagation matrix of each node in the process of transmitting transaction information. Based on the loss coefficient, the invention can accurately reflect the loss of transaction information in the transmission process, which is conducive to feedback the real situation of the information integrity of the online transaction platform and provides more accurate and comprehensive information support for the transaction platform.

[0016] Optionally, the information feature analysis model is established according to the platform transaction information, and the transaction information transmission result is analyzed by using the information feature analysis model to obtain the transaction feature information of the online transaction platform, including: combining the transaction risk control system structure and the platform transaction information to construct an information feature analysis model; analyzing the transaction information transmission result by using the information feature analysis model to obtain the transaction feature information of the online transaction platform. The present invention conducts an in-depth analysis of the transaction information transmission result to reveal the inherent laws and data set characteristics of the transaction behavior, which helps the transaction platform to better grasp user needs and market dynamics, thereby providing a more transparent and fairer transaction environment and enhancing the trust of the online transaction platform.

[0017] Optionally, the information feature analysis model satisfies the following relationship:

[0018]

[0019] Among them, Z represents the transaction feature information after analysis, t 1 Represents the first feature information in the feature label model, σ 1 represents the weight corresponding to the first feature, H 1* The first characteristic information contained in the delivery matrix representing the transaction information, t 2 Represents the second feature information in the feature label model, σ 1represents the weight corresponding to the second feature, H 2* The second characteristic information contained in the delivery matrix representing the transaction information, t m represents the mth feature information in the feature label model, σ m represents the weight corresponding to the mth feature, H m* The mth characteristic information contained in the delivery matrix representing the transaction information. The model of the present invention can integrate multiple characteristic information to evaluate the overall characteristics of the transaction, ensure the influence of important characteristic information on the result, and can better adapt to the complexity and diversity of transaction information.

[0020] Optionally, the analysis of the platform transaction status based on the transaction feature information and the platform transaction information to ensure the risk control and safe operation of the online transaction platform includes: setting online transaction scheduling indicators based on historical transaction information, the online transaction scheduling indicators including the correlation status, timing status and feature label status of online transactions; analyzing the platform transaction status in combination with the transaction feature information, the platform transaction information and the online transaction scheduling indicators, and obtaining the operation status of the online transaction platform; adjusting the online transaction platform based on the operation status of the online transaction platform to achieve risk control and safe operation of the online transaction platform. The present invention combines transaction feature information and platform transaction information to more accurately identify potential risk points in transactions, wherein the online transaction scheduling indicators make risk identification more comprehensive and specific, and help promote the platform to provide more convenient, efficient and secure transaction services.

[0021] In the second aspect, in order to efficiently execute a transaction risk control method provided by the present invention, the present invention also provides a transaction risk control system, the system includes a data collection module, a data transmission and collation module, a transaction data analysis module and an intelligent trading module, the data collection module, the data transmission and collation module, the transaction data analysis module and the intelligent trading module are interconnected, and the system executes the transaction risk control method as described in the first aspect of the present invention. The system has a compact structure and stable performance, and can stably execute the transaction risk control method provided by the present invention, thereby improving the overall applicability and practical application capabilities of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0023] Figure 1 This is a flow chart of the transaction risk control method of the present invention;

[0024] Figure 2 It is a flow chart of a transaction information integration model of the transaction risk control method of the present invention;

[0025] Figure 3 A schematic diagram of the analysis process of transaction feature information of the transaction risk control method of the present invention;

[0026] Figure 4 This is a structural diagram of the transaction risk control system of the present invention. DETAILED DESCRIPTION

[0027] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that these specific details do not need to be adopted to implement the present invention. In other examples, in order to avoid confusing the present invention, known circuits, software or methods are not specifically described.

[0028] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or subcombination. In addition, it should be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale.

[0029] See also Figure 1 The present invention provides a transaction risk control method. The implementation steps and specific process of the transaction risk control method are as follows:

[0030] S1. Obtain platform transaction information through the transaction risk control system. The specific steps and implementation contents are as follows:

[0031] In the embodiment, a data collection module is set in the transaction risk control system, and the platform transaction information is obtained by using the data collection module. In the embodiment, the online transaction platform information data of Shenzhen Xiaofu Technology Co., Ltd. is taken as an example.

[0032] In an optional embodiment, the data sources and data collection processes of the online trading platform of Shenzhen Xiaofu Technology Co., Ltd. are carefully analyzed and planned, and a comprehensive data collection module is constructed. The above collection module not only covers the core data sources of the internal system of the trading platform, but can also receive external information sources, such as payment gateway transaction records, market analysis reports provided by third-party data service providers, etc., to ensure the comprehensiveness and multi-dimensional coverage of the platform transaction information.

[0033] In order to efficiently and accurately collect platform transaction information, the company deployed a data collection module in the corresponding online trading platform, and configured the corresponding device parameters and collection methods according to the characteristics of each data source. The above collection methods include but are not limited to real-time calls through the API interface to obtain the latest transaction data, deep mining of historical transaction records using database direct connection technology, and capturing user behavior details through log analysis technology. The comprehensive application of the above data collection strategies has greatly improved the collection efficiency and stability of the data collection module.

[0034] At the same time, in order to ensure the security and maintainability of transaction information, a data storage solution is also configured. The above solution not only involves the selection of databases to ensure high concurrent reading and writing capabilities and long-term scalability of data, but also goes deep into the design of data table structures, striving to improve the efficiency of data query and analysis through reasonable table structure arrangement and index optimization. In addition, multiple data backup and recovery strategies are adopted to prevent data loss or damage, providing a solid guarantee for the long-term preservation and reliable use of data.

[0035] In the embodiment, based on the online trading platform of Shenzhen Xiaofu Technology Co., Ltd., a series of measures were taken to clarify the data source, optimize the data collection process, design a reasonable data storage solution, etc., and successfully build an efficient, stable and secure data collection module, which provides strong support for the stable operation and precise risk control of the online trading platform.

[0036] Furthermore, the setting method of the data collection module in the embodiment is only an optional condition of the present invention. In one or some other embodiments, the system data collection module can be adjusted and optimized according to the data collection requirements and the actual situation of the trading platform. Different trading platforms have different data types, data volumes and data frequencies. The data collection module can be adjusted according to actual needs, so that the transaction risk control system can more flexibly adapt to various complex and changeable business scenarios, ensuring the accuracy and timeliness of data collection.

[0037] S2. Set up a transaction data transmission and sorting mechanism to sort and detect platform transaction information and obtain transaction information transmission results. The specific steps and implementation contents are as follows:

[0038] First, a data transmission and sorting module is set in the transaction risk control system. The data transmission and sorting module sets a transaction data transmission and sorting mechanism based on the platform transaction information. The transaction data transmission and sorting mechanism set in the embodiment includes a transmission node transaction information model, a transaction data loss degree analysis model and a transaction information integration model. The specific contents are as follows:

[0039] In an optional embodiment, the transmission node transaction information model is used to obtain the propagation matrix of any node in the transaction information transmission process, and the specific content is as follows:

[0040] The embodiment optimizes the existing graph convolutional neural network and sets up a transaction data transmission and sorting mechanism, which means that a transmission node transaction information model, a transaction data loss analysis model and a transaction information integration model are established in the data transmission and sorting module based on the platform transaction information.

[0041] In this embodiment, the convolutional neural network is further optimized. As a deep learning model for structural data, the optimized convolutional neural network can extend the core concept of the original convolutional neural network to a more complex and flexible information structure, so that the optimized convolutional neural network can accurately capture and analyze the complex interactive relationship between different nodes of information.

[0042] That is, the optimized convolutional neural network overturns the simple method of treating signals as discrete points in Euclidean space in traditional signal technology, and instead treats information signals as attributes or features on nodes. A new node information update mechanism is introduced in the embodiment, that is, by aggregating the information of each node and neighboring nodes, the information representation of different nodes is dynamically adjusted and optimized, which not only reflects the global topological characteristics of the optimized convolutional neural network, but also deeply reveals the potential interactions and influence relationships between different information nodes.

[0043] Unlike traditional neural networks, due to the connection and mutual influence between different nodes, the propagation of gradients between nodes becomes more complex and flexible, and it is necessary to consider the weight matrix, bias term, and matrix information corresponding to the same node. When the model parameters are updated, the above parameters can be uniformly adjusted according to the gradients and situations of different nodes, thereby realizing information processing and classification of different nodes on irregular structured data.

[0044] The transaction information model of the transport node in the data transmission and sorting module in the embodiment satisfies the following relationship:

[0045]

[0046] Among them, H irepresents the propagation matrix of each node in the process of transaction information transmission, softmax represents the Softmax function, Y i represents the adjacency matrix of the ith node, B i represents the degree matrix of the i-th node matrix, S i (t) represents the input feature matrix corresponding to the timestamp of the i-th node, Q i (t) represents the weight matrix of the i-th node, P i (t) represents the bias vector corresponding to the timestamp of the i-th node

[0047] The propagation matrix of each node in the process of transaction information transmission can describe the model or framework of the information propagation characteristics of any node. It mainly includes but is not limited to the connection relationship between different nodes, the efficiency, direction and path of information propagation, security and reliability, etc. Among them, the performance indicators such as the speed, weight and timestamp of information propagation of different nodes will affect the real-time and accuracy of transaction information.

[0048] The adjacency matrix of the i-th node, that is, the matrix corresponding to the i-1-th node. The elements of the above adjacency matrix represent the connection status between the nodes. For a transaction information with n nodes, the adjacency matrix is ​​an n×n matrix. If there is an edge between node i and node j, the element in the i-th row and j-th column or the element in the j-th row and i-th column in the matrix depends on whether it is a directed graph or an undirected graph. The value is 1 or the weight of the edge; if there is no edge, the value of the corresponding position is 0.

[0049] The adjacency matrix of the i-th node above satisfies the following relationship:

[0050] Y i =h i-1 +b

[0051] Among them, Y i represents the adjacency matrix of the ith node, h i+1 represents the matrix of the i-1th node, and b represents the preset unit matrix.

[0052] The preset unit matrix can also be called the unit matrix, the identity matrix, or the standard unit matrix, which refers to an n×n matrix in which the elements on the main diagonal are all 1, that is, the element in the i-th row and i-th column is 1, and the remaining elements are all 0. In the embodiment, the unit matrix can be used to represent the starting point or unit state form of the transformation.

[0053] The degree matrix is ​​a special matrix that can be used to represent the degree of each node. The above matrix is ​​a square matrix with the number of rows and columns equal to the number of nodes in the graph. The degree matrix is ​​a diagonal matrix, that is, except for the diagonal elements, all other elements in the matrix are 0.

[0054] For each diagonal element in the degree matrix, it represents the degree of the corresponding node. In an undirected graph, this degree is the number of edges connected to the node; in a directed graph, the degree can be further subdivided into out-degree and in-degree. Since the degree matrix is ​​a diagonal matrix, except for the diagonal elements, all other elements in the matrix are 0, which means that the degree or number of edges between nodes will not directly affect the value of the non-diagonal elements.

[0055] In an optional embodiment, a degree matrix of any node matrix is ​​established, and the specific contents are as follows:

[0056] First, we need to determine the degree of each node in the graph; then, we create a square matrix equal to the number of nodes, set the diagonal elements to the degree of the corresponding node, and set the remaining elements to 0.

[0057] If the adjacency matrix of the above i-th node matrix satisfies the following relationship:

[0058]

[0059] The degrees of the above nodes are as follows: the degree of node 1 is 1; the degree of node 2 is 3; the degree of node 3 is 1; the degree of node 4 is 1.

[0060] Therefore, the degree matrix of the i-th node matrix can be expressed as:

[0061]

[0062] The degree matrix is ​​a diagonal matrix used to represent the degree information of different nodes. It is constructed based on the degree or in-degree of each node in the graph, and all non-diagonal elements are 0.

[0063] The parameters such as the propagation matrix, degree matrix and feature matrix constructed in the embodiment have each row representing an observation at a time point, and each column representing a specific attribute or feature. At the same time, the matrix can contain static features such as the inherent attributes of the node and dynamic features that change over time as needed.

[0064] Analyze the transaction information matrix based on the time window. In order to capture the time dependency and short-term trends in the time series, the information matrix can be analyzed by time window, including calculating statistics such as mean, median, standard deviation, maximum, minimum, etc. within a given time window. The above statistics can reflect the overall characteristics of the data within a specific time window. In addition, the sliding window technology can be applied to gradually move the time window to analyze the dynamic changes of the entire time series.

[0065] In this embodiment, a time series model is introduced. In order to better understand the complexity of time series data and predict future trends, a time series analysis model is used in the embodiment, including but not limited to ARIMA, seasonal decomposition, exponential smoothing, etc. The above model not only considers the statistics within the time window, but also uses the patterns and periodicity in historical data to predict future values.

[0066] Then, sequence-to-sequence conversion is performed based on the time series model. For scenarios that need to consider both past and future information, time series data can be converted to sequence-to-sequence format and processed using deep learning models such as recurrent neural networks, long short-term memory networks, or gated recurrent units, which can learn long-term dependencies in time series and generate predicted or converted sequences.

[0067] Furthermore, multidimensional time series analysis can be performed. If the node information contains multiple related time series, multidimensional time series analysis can be performed. It is necessary to consider the synchronization and asynchrony between different sequences, as well as the interaction between different multidimensional time series. Covariance analysis, principal component analysis (PCA) or tensor decomposition can be used to analyze the complex relationship between different time series.

[0068] When calculating the weight matrices of different nodes, we first need to clarify the meaning and purpose of the weight matrix in a specific context or application. Since the definition of the weight matrix can vary from application to application, it can be combined with the number or frequency of interactions between different nodes. The weight matrix elements can represent the weight from node i to node j at time t, that is, the number of interactions between nodes i and j before time t.

[0069] The bias vectors corresponding to different nodes at timestamps can adjust the function output or linear transformation results in the model, allowing the model to have a non-zero output even when there is no input. In the data transmission and sorting module, the bias term can be regarded as the threshold adjustment of the function for the node, allowing the node to be activated when the input is zero, if the bias term is large enough and the function allows it.

[0070] In this embodiment, a neural network framework is used and the bias vector of the initialization node is designed. The specific calculation content is as follows:

[0071] import tensorflow as tf

[0072] #Assume that there are n transaction data nodes in this embodiment

[0073] num_neurons = n

[0074] # Initialize the bias vector with zeros

[0075] bias=tf.Variable(tf.zeros([num_neurons]),name='bias')

[0076] #Combine the bias vector with the weight of any node

[0077] #output=tf.matmul(input,weights)+bias,output=(P i (t))

[0078] In another optional embodiment, the loss situation of platform transaction information is obtained by using a transaction data loss analysis model, and the specific contents are as follows:

[0079] In the embodiment, the sample matrix and the actual matrix are compared and analyzed. The transaction data loss analysis function is an effective similarity measurement technology that can accurately reflect the internal relationship between different information. The model can automatically distinguish the similarity of transaction behaviors, patterns or characteristics, which is helpful to identify transaction fraud detection, market trend prediction and other information.

[0080] The transaction data loss degree analysis function trains the network by minimizing the distance between similar samples and maximizing the distance between dissimilar samples. A threshold is defined based on historical information and rules. When the distance between two similar samples is less than the preset threshold, they are considered to be the same; when the distance between two dissimilar samples is greater than the preset threshold, they are considered to be different. The above transaction data loss degree analysis model satisfies the following relationship:

[0081]

[0082] Among them, LOSS represents the loss coefficient of transaction information, N represents the total number of nodes of transaction information, η represents the similarity coefficient between the propagation matrix of the i-th node and the sample propagation matrix, l represents the distance between the propagation matrix of the i-th node and the sample propagation matrix, and α represents the propagation matrix distance threshold.

[0083] The similarity coefficient between different node propagation matrices and sample propagation matrices refers to a quantitative indicator that measures the similarity between the two matrices in terms of structure, element distribution or a certain specific property. In addition, for the similarity coefficient between different node propagation matrices and sample propagation matrices, if they have similar structures or can be converted into similar representations such as vectors or sets, any one or more of the above similarity measures can be used to calculate. However, if their structures are very different or cannot be directly compared, they need to be preprocessed or converted first so that an appropriate similarity measure can be applied.

[0084] The similarity coefficients between the above different node propagation matrices and the sample propagation matrices satisfy the following relationship:

[0085]

[0086] Among them, η i represents the similarity coefficient between the propagation matrix of the i-th node and the propagation matrix of the sample, δ represents the scaling parameter of the feature vector, T(H i ) represents the eigenvector of the propagation matrix mapping of the i-th node, H c The eigenvector representing the sample propagation matrix mapping.

[0087] The scaling parameter of an eigenvector refers to the way in which the eigenvalue can describe the effect of a linear transformation such as matrix multiplication on a vector. When a linear transformation such as matrix A acts on a non-zero vector B, if the resulting vector AB is collinear with B, that is, they are on the same straight line but may be in opposite directions or have different lengths, then B is called an eigenvector of A, and the scaling factor δ between AB and B is the corresponding eigenvalue. The above scaling factor δ is actually the scaling parameter of the eigenvector B under the transformation.

[0088] The eigenvector of matrix mapping refers to a non-zero vector whose direction remains unchanged or only undergoes scaling transformation without rotation under the linear transformation of the matrix. The eigenvector maintains the same direction or only reverses direction under the transformation of the matrix, making the correlation vector the key to understanding the sample propagation vector.

[0089] In another optional embodiment, a transaction information integration model is constructed based on the delivery node transaction information model and the transaction data loss degree analysis model, and the delivery information set of the online transaction platform is obtained using the transaction information integration model. The specific contents are as follows:

[0090] In the process of transmitting transaction information, the propagation matrix describes the propagation mode and efficiency of transaction information from one node to other nodes, including parameters such as the connection strength between different nodes, information propagation speed, and reliability. At the same time, in the process of transmitting transaction information, due to factors such as network failure, node failure, and malicious attacks, transaction data may be lost or damaged. The data loss results can be quantified as the amount of lost data, the value of lost data, and the degree of impact on the transaction process.

[0091] The transaction information transmission matrix combines the propagation matrix and the transaction data loss results to comprehensively describe the status, efficiency and risk of transaction information during the transmission process. It not only considers the propagation mode and efficiency of transaction information in the data transmission and sorting module, but also considers the possible loss of transaction information during the transmission process.

[0092] The above transaction information integration model satisfies the following relationship:

[0093]

[0094] Among them, H represents the transmission matrix of transaction information, LOSS represents the loss coefficient of transaction information, N represents the total number of nodes of transaction information, and H i Represents the propagation matrix of each node in the process of transaction information transmission.

[0095] In an optional embodiment, the transaction information integration model obtains the process of the delivery matrix of transaction information, see Figure 2 , where H1, H2, H3, H4, H5, H6, H7, and Hi represent the propagation matrices corresponding to different nodes in the process of transaction information transmission.

[0096] based on Figure 2 It can be seen that the transaction data transmission and sorting mechanism established by the above-mentioned data transmission and sorting module mainly analyzes the connection relationship and information flow of different nodes through the transmission node transaction information model, and obtains the propagation matrix of different nodes in the process of transaction information transmission. The above-mentioned matrix contains parameters such as connection strength, information propagation speed, and reliability between different nodes. Then, the transaction data loss degree analysis model is used to conduct risk assessment and loss identification on the transaction information of the system, and analyze the possible losses of transaction information during the transmission process, such as data loss or damage caused by network failure, node failure, malicious attack, etc., and then the probability of transaction information loss, the amount of lost data, the value of lost data and the degree of impact on the transaction process can be quantified. Finally, the transaction information propagation matrix and the data loss risk assessment results are combined to comprehensively construct a transaction information integration model and obtain a transaction information set, conduct in-depth analysis on the transaction information of the system, identify information transmission information and optimization space, and based on the transaction information integration results, the reliability and timeliness of transaction information can be more accurately evaluated, thereby making more reasonable transaction decisions, which will help reduce transaction risks and improve transaction success rates.

[0097] In the embodiment, incorporating the loss coefficient into the transaction information integration model can more comprehensively evaluate the risks that transaction information may suffer during the transmission process, help to better understand potential data loss or damage, and thus formulate more effective risk management strategies.

[0098] Furthermore, the method for acquiring the transaction information set in the present embodiment is only an optional condition of the present embodiment. In one or some other embodiments, the method for acquiring the transaction information set can be optimized according to the transaction information requirements and the transaction risk control system conditions. Different transaction scenarios and risk control systems have different requirements for transaction information. Optimizing the acquisition method can enable the model to meet the information acquisition requirements in different scenarios and improve the versatility and extensibility of the model.

[0099] S3. Establish an information feature analysis model based on the platform transaction information, and use the information feature analysis model to analyze the transaction information transmission results to obtain the transaction feature information of the online transaction platform. The specific steps and implementation contents are as follows:

[0100] An information feature analysis model is constructed by combining the transaction risk control system structure and platform transaction information. The transaction information transmission results are analyzed through the information feature analysis model to obtain the transaction feature information of the online trading platform. The specific contents are as follows:

[0101] The transaction data analysis module first gathers and absorbs the transaction information set from the data transmission and collation module, and then uses the advanced algorithms and models in the analysis module to conduct in-depth analysis and deconstruction of the transaction information set to extract key transaction characteristics, trends and potential risk points, which helps to ultimately form a detailed and effective transaction information analysis report.

[0102] In order to more accurately grasp transaction needs and behavior patterns, the transaction data analysis module is further used to sort out and analyze the relationships among customer information, capital flow trajectories, transaction attributes, etc. in the transaction information set. By constructing a multi-dimensional and three-dimensional customer transaction information portrait, it can not only clearly depict the basic information such as the financial status and credit rating of the transaction participants, but also deeply understand their psychological characteristics such as trading habits and risk preferences, providing a solid data foundation for personalized services and risk prevention and control.

[0103] The transaction data in the embodiment has the characteristics of large scale, diversity and high circulation. The system deploys an efficient data transmission and sorting module, which can quickly capture and aggregate transaction data information from various business systems. It can also convert it into a transaction information set with clear structure and complete content through a series of operations such as cleaning, conversion and integration. The above process not only realizes seamless connection and information sharing between internal information systems, but also lays the foundation for subsequent transaction information analysis.

[0104] On this basis, the transaction data analysis module is used to integrate and analyze transaction information. On the other hand, external data sources are introduced. The above external data has certain accuracy, reliability and stability, which can become an important way to supplement internal transaction information, making up for the lack of transaction data while improving the integrity of transaction information. The transaction data analysis module integrates and complements internal and external data, effectively solving information quality problems such as information asymmetry, and providing a more comprehensive and accurate information basis for transaction risk control methods.

[0105] The information feature analysis model in the transaction data analysis module needs to satisfy the following relationship:

[0106]

[0107] Among them, Z represents the transaction feature information after analysis, t 1 Represents the first feature information in the feature label model, σ 1 represents the weight corresponding to the first feature, H 1* The first characteristic information contained in the delivery matrix representing the transaction information, t 2 Represents the second feature information in the feature label model, σ 1 represents the weight corresponding to the second feature, H 2* The second characteristic information contained in the delivery matrix representing the transaction information, t m represents the mth feature information in the feature label model, σ m represents the weight corresponding to the mth feature, H m* The mth feature information contained in the delivery matrix representing the transaction information.

[0108] In an optional embodiment, the transaction information set processed by the transaction data analysis module is not limited to the direct transaction data of the transaction risk control system, but also covers the management data recorded in the system, which reflects the operation status and decision-making trajectory of the enterprise. In addition, the key indicator data of operation management is also included in the transaction data analysis module, thereby providing an important basis for the evaluation of business performance and monitoring operation efficiency.

[0109] Based on the above transaction data analysis module, it can be seen that this module also integrates professional data resources from third parties and relevant transaction information on the Internet. Third-party data is professional and targeted, and can provide a unique perspective and in-depth insights for transaction analysis; while Internet data is real-time, extensive and diverse, providing rich information resources for capturing transaction market dynamics and consumer behavior analysis.

[0110] In the embodiment, the transaction data analysis module can fully integrate business system transaction data, management system management data, operation management indicator data, third-party professional data and Internet transaction information, thereby constructing a multi-dimensional and multi-level data analysis framework. The above framework not only enhances the breadth and depth of data analysis, but also improves the accuracy and timeliness of analysis, and provides strong data support for risk management, business decision-making and strategic planning of the transaction risk control system.

[0111] When analyzing the data characteristics of transaction information sets, the transaction data analysis module adopts a series of technical means and strategies to explore the potential information and interrelationships of transaction data. In view of the diversity and complexity of transaction data, a flexible data extraction and conversion process is designed to ensure that the data can be processed and analyzed efficiently and accurately.

[0112] First, advanced data analysis tools are used to extract structured data targets from transaction information sets. For non-standard log-type and library table data, special scripts are written to ensure that the above information can be smoothly extracted and stored in the Hadoop distributed file system, laying the foundation for subsequent transaction data analysis.

[0113] For the acquired structured data, it is directly converted into Hive table format, and a series of pre-processing tasks such as data cleaning, fusion, and conversion are completed based on Hive's powerful data processing capabilities. At the same time, a knowledge graph is also constructed to graphically display the complex relationship between data, providing strong support for subsequent in-depth analysis. In the entire process of data processing, Spark's parallel computing capabilities are fully utilized, which significantly improves the analysis efficiency and effect of system transaction data.

[0114] In the transaction data analysis module, a transaction data feature label model is constructed. In the embodiment, statistical principles and methods are used to conduct in-depth statistical analysis on the transaction information set of the transaction platform. By calculating statistics such as variance and standard deviation, the key features of different data items can be extracted to form basic data elements. The above basic data elements are used as the cornerstone of modeling and analysis and are substituted into the iterative calculation process to continuously optimize and improve the feature label model.

[0115] In an optional embodiment, the construction of a multi-dimensional tag library for customers is to complete the classification, extraction and processing of different characteristic information in the transaction information. Shenzhen Xiaofu Technology Co., Ltd. is the center, and various data resources inside and outside the enterprise are integrated to form a rich and detailed transaction information set by retrieving and analyzing the basic information, equity structure, personnel changes, business conditions, industrial and commercial registration, legal cases, credit records, guarantees, administrative penalties, social public opinion and other multi-dimensional information of the enterprise. The above transaction information set not only covers the static attributes of the enterprise, but also includes dynamic change indicators, such as the number of overdue payments, etc., which can provide a full-dimensional and multi-angle corporate portrait.

[0116] In the process of establishing the feature label model, a variety of unsupervised learning algorithms can be used, such as cluster analysis, association rule mining, collaborative filtering, etc., to achieve in-depth insights and accurate classification of online transaction behaviors. The above models not only help to identify value groups in the transaction process, but also effectively predict potential risks and provide strong support for corporate operational decisions.

[0117] In addition, we also make full use of advanced machine learning algorithms such as GDBT, SVM, RF, LR, ARIMA and XGBoost, combined with big data technology, to analyze the transaction customers in the transaction information set, including but not limited to in-depth analysis and mining of the basic attributes and behavioral characteristics of the transaction participants. Based on this, we can successfully build a group portrait of transaction customers, a product preference model, and an honesty prediction model, which will provide a scientific basis for the online trading platform's precision marketing, risk management and other goals.

[0118] This embodiment also introduces support vector machine (SVM) into the model library. By continuously optimizing model parameters and algorithm design, it helps to continuously improve the performance of support vector machine (SVM) and other models to meet the business needs and risk prevention and control of the enterprise online trading platform.

[0119] The transaction feature information after analysis satisfies the following relationship:

[0120]

[0121] Among them, the feature information in the feature label model includes, but is not limited to, basic transaction information such as transaction time, amount, type, and status; transaction user information includes transaction user ID, account information, browsing history, purchase history, click behavior, etc.; commodity or service information such as commodity or service ID, name, category, price, inventory; abnormal transaction identification such as abnormal transaction amount, abnormal transaction time, etc., used to identify potential fraud; features in transaction time series analysis, such as transaction frequency, transaction interval, transaction trend, etc., are used to analyze user transaction habits and patterns. In the embodiment, transaction feature information plays an important role in online transaction data analysis, which helps to understand transaction needs, optimize product recommendations, improve user experience, identify potential risks, etc. At the same time, through in-depth mining and analysis of relevant feature information, more accurate marketing decision support can also be provided to enterprises.

[0122] See also Figure 3 , the transaction feature information after analysis satisfies the following schematic diagram, where 1, 2, and m represent the first, second, and mth feature information in the feature label model, respectively, and δ 1 , δ 2 , δ m They respectively represent the weights corresponding to different feature information, H1*, H2*, Hm* respectively represent different feature information, and t1, t2, tm respectively represent different feature information in the feature label model after analysis.

[0123] Based on the schematic diagram, it can be seen that the transaction feature information after analysis includes all feature information in the feature label model, and the weights corresponding to different feature information are combined to analyze and process the transaction feature information of the system. The above feature information can be replaced and supplemented based on the key parameters of the platform transaction information and the structural characteristics of the platform transaction system. The embodiment comprehensively covers all feature information in the feature label model, and analyzes it in combination with the corresponding weights, which can more accurately capture the nature and trend of transaction behavior, help identify potential transaction patterns, abnormal behaviors or market trends, and thus improve the accuracy of transaction decisions.

[0124] The transaction data analysis module analyzes transaction feature information based on big data and machine learning technology, which can provide strong information support for the intelligent trading module. By automatically analyzing and processing massive transaction data, it helps the intelligent trading module to quickly generate decision-making recommendations or execute transaction instructions, thereby effectively ensuring the risk control and safe operation of the online trading platform.

[0125] Furthermore, the transaction information analysis technology and specific steps in the present embodiment are merely optional conditions of the present embodiment. In one or some other embodiments, the transaction information analysis technology may be adjusted according to the actual operation of the system and the transaction information analysis requirements. By adjusting the transaction information analysis technology, the analysis process may be continuously optimized, unnecessary computing overhead may be reduced, and the analysis speed and accuracy may be improved, thereby enhancing the overall performance and analysis efficiency of the system.

[0126] S4. Analyze the platform transaction status based on transaction feature information and platform transaction information to ensure risk control and safe operation of the online trading platform. The specific steps and implementation contents are as follows:

[0127] First, online transaction scheduling indicators are set based on historical transaction information. In the embodiment, online transaction scheduling indicators mainly include the association status, timing status and feature label status of online transactions. Furthermore, in the embodiment, a transaction strategy scheduling device, an intelligent display device and a data storage device are provided in the intelligent transaction module. The transaction strategy scheduling device, the intelligent display device and the data storage device realize transaction risk control according to the transaction data analysis results.

[0128] As the risk prevention and control analysis center of the online trading platform, the transaction strategy scheduling device brings together diversified data information from the data collection module, data transmission and sorting module, and transaction data analysis module, including but not limited to transaction information structure relationships, external third-party information, internal company information, and extensive external data such as industrial and commercial registration, judicial records, real estate information, provident fund payment, social security status, etc. By using advanced machine learning algorithms such as support vector machines, carefully constructing and continuously optimizing risk control and credit models, the above process not only realizes real-time risk monitoring of the entire chain of online trading business, but also ensures the accuracy and fairness of credit ratings, building a solid line of defense for the security of online transactions.

[0129] First, correlation analysis is performed to obtain the correlation status of online transaction information.

[0130] The intelligent trading module has pre-set transaction data association rules. By using unsupervised learning technology, the transaction strategy scheduling device can automatically and deeply explore the complex relationships and internal laws hidden in the transaction data. The above association rules are not only highly practical and intuitive, but also can reveal the connection between various related factors, thus laying a solid information foundation for the early prediction and effective prevention and control of transaction risks. The above process not only enhances the intelligence level of transaction decision-making, but also significantly improves the accuracy and efficiency of risk management.

[0131] Second, perform timing analysis to obtain the timing of online transaction information.

[0132] Next is the time series analysis, which focuses on the changes and differences in the time series of trading data. It can use a variety of statistical methods such as autocorrelation analysis and spectral analysis, and combine the construction and verification of statistical models to deeply analyze the trends and periodic laws contained in historical trading data. The above process not only increases the online trading platform's ability to predict future market dynamics, but also shows immeasurable practical value in optimizing prediction models, formulating effective control strategies, and implementing efficient data filtering. The time series analysis provides a solid data foundation for the trading decision-making process, ensuring the scientificity and foresight of trading risk decisions.

[0133] Third, label analysis is performed to obtain characteristic labels of online transaction information.

[0134] Then, feature tag analysis is performed. This step aims to deeply analyze the feature tag situation of online transaction information through tag aggregation technology. The core of tag analysis is to use advanced clustering and statistical analysis methods to finely process specific feature data or identification information involved in the transaction. In particular, for high-risk feature information, the system can quickly identify customer groups with similar risk characteristics and accurately assess their potential risk probability.

[0135] The above process not only greatly simplifies the risk management and analysis process, but also significantly improves the response speed and accuracy of risk prediction. Through the real-time generation of risk tags and prediction reports, enterprise risk managers and operational decision makers can quickly obtain key information, thereby formulating more accurate and efficient risk response strategies. Tag analysis not only provides strong technical support for corporate risk management, but also provides a solid guarantee for enterprises to cope with the ever-changing market environment and ensure the security and stability of transactions.

[0136] Then, the platform transaction status is analyzed by combining the transaction feature information, platform transaction information and online transaction scheduling indicators, and the operation status of the online transaction platform is obtained.

[0137] After completing the correlation analysis, time series analysis and label analysis, the intelligent trading module will further integrate the transaction feature information, platform transaction information and online transaction scheduling indicators to achieve a comprehensive and multi-dimensional analysis of the platform transaction status. The above process combines the deep mining and real-time monitoring of transaction data, which can reveal the internal logic and external performance of online transaction activities, and thus obtain a comprehensive insight into the operation of the online trading platform.

[0138] Furthermore, the transaction risk control system of this embodiment will conduct a comprehensive evaluation and comparative analysis based on key elements in the transaction feature information, such as transaction frequency, amount distribution, user behavior patterns, etc., combined with key indicators such as order processing efficiency, payment success rate, user satisfaction in the platform transaction information, and real-time data such as transaction flow, system load, response time in online transaction scheduling indicators.

[0139] Through the above series of complex analysis steps, the intelligent trading module can accurately judge the operational health of the platform and identify potential trading bottlenecks, risk points or optimization space. At the same time, the system can also predict future trading trends and provide forward-looking decision-making support for platform managers, helping them to flexibly adjust operating strategies and optimize resource allocation, thereby ensuring the continued stable operation and business growth of the online trading platform.

[0140] Finally, adjustments are made to the online trading platform based on its operating conditions to achieve risk control and safe operation of the online trading platform.

[0141] Finally, based on the above-mentioned in-depth analysis and comprehensive assessment of the operation of the online trading platform, the intelligent trading module takes a series of targeted adjustment measures to help achieve risk control and safe operation of the online trading platform.

[0142] The adjustment measures in the embodiment include but are not limited to: optimizing transaction strategies, adjusting transaction rules, rate structures or preferential strategies according to transaction feature information and market dynamics, so as to balance user experience and platform revenue, while reducing the probability of fraud and risky transactions.

[0143] Enhanced risk control: For high-risk customers or behavior patterns identified in label analysis, stricter review processes are implemented, transaction amounts are restricted, or other risk control measures are taken to effectively control credit risk and operational risk.

[0144] Improve system performance Based on the system load and response time reflected by online transaction scheduling indicators, optimize server configuration, upgrade software versions or introduce more efficient algorithms to increase transaction processing speed and stability and reduce the risk of system failure and data loss.

[0145] In addition, it is necessary to improve supervision and compliance, ensure that the platform operations comply with relevant laws, regulations and regulatory requirements, strengthen the management of data protection, privacy policy and anti-money laundering, and improve the compliance and credibility of the platform; through user feedback and market research, continuously optimize the platform interface, transaction process and customer service, improve user satisfaction and loyalty, and lay a solid foundation for the long-term development of the platform.

[0146] Based on the transaction risk control system of the present invention, the operation of the online transaction platform is adjusted and optimized, and the intelligent transaction module can effectively achieve the goals of risk control and safe operation, and ensure the stable operation and sustainable development of the online transaction platform.

[0147] See also Figure 4 In an optional embodiment, in order to efficiently execute the transaction risk control method provided by the present invention, the present invention also provides a transaction risk control system, which includes a data collection module, a data transmission and sorting module, a transaction data analysis module and an intelligent trading module. The data collection module, the data transmission and sorting module, the transaction data analysis module and the intelligent trading module are interconnected. The system executes the specific steps of the relevant embodiments of the transaction risk control method provided by the present invention. The transaction risk control system of the present invention has a complete structure, is objective and stable, and can efficiently execute the transaction risk control method of the present invention, thereby improving the overall applicability and practical application capabilities of the present invention.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

[0149] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A transaction risk control method, characterized in that: The steps include: Obtain platform transaction information through the transaction risk control system; Setting up a transaction data transmission and sorting mechanism to sort and detect the transaction information of the platform and obtain the transaction information transmission results; Establishing an information feature analysis model based on the platform transaction information, and using the information feature analysis model to analyze the transaction information transmission results to obtain transaction feature information of the online transaction platform; The platform transaction status is analyzed based on the transaction feature information and the platform transaction information to ensure risk control and safe operation of the online transaction platform.

2. The transaction risk control method according to claim 1, characterized in that: The setting of the transaction data transmission and sorting mechanism to sort and detect the platform transaction information and obtain the transaction information transmission result includes: A transaction data transmission and sorting mechanism is set up according to the platform transaction information, and the transaction data transmission and sorting mechanism includes a transmission node transaction information model, a transaction data loss analysis model and a transaction information integration model.

3. The transaction risk control method according to claim 2, characterized in that: The setting of the transaction data transmission and sorting mechanism to sort and detect the platform transaction information includes: Using the transmission node transaction information model, a propagation matrix of any node in the transaction information transmission process is obtained; Using the transaction data loss analysis model to obtain the loss of platform transaction information; A transaction information integration model is constructed based on the delivery node transaction information model and the transaction data loss degree analysis model, and the delivery information set of the online transaction platform is obtained using the transaction information integration model.

4. The transaction risk control method according to claim 2, characterized in that: The transport node transaction information model satisfies the following relationship: Among them, H i represents the propagation matrix of each node in the process of transaction information transmission, softmax represents the Softmax function, h i represents the matrix of the i-th node, Y i represents the adjacency matrix of the ith node, B i represents the degree matrix of the i-th node matrix, S i (t) represents the input feature matrix corresponding to the timestamp of the i-th node, Q i (t) represents the weight matrix of the i-th node, P i (t) represents the bias vector corresponding to the timestamp of the i-th node.

5. The transaction risk control method according to claim 3, characterized in that: The transaction data loss degree analysis model satisfies the following relationship: Among them, LOSS represents the loss coefficient of transaction information, N represents the total number of nodes of transaction information, η represents the similarity coefficient between the propagation matrix of the i-th node and the sample propagation matrix, l represents the distance between the propagation matrix of the i-th node and the sample propagation matrix, and α represents the propagation matrix distance threshold.

6. The transaction risk control method according to claim 2, characterized in that: The transaction information integration model satisfies the following relationship: Among them, H represents the transmission matrix of transaction information, LOSS represents the loss coefficient of transaction information, N represents the total number of nodes of transaction information, and H i Represents the propagation matrix of each node in the process of transaction information transmission.

7. The transaction risk control method according to claim 1, characterized in that: The step of establishing an information feature analysis model based on the platform transaction information and analyzing the transaction information transmission results using the information feature analysis model to obtain the transaction feature information of the online transaction platform includes: Constructing an information feature analysis model based on the transaction risk control system structure and the platform transaction information; The transaction information transmission results are analyzed through the information feature analysis model to obtain the transaction feature information of the online transaction platform.

8. The transaction risk control method according to claim 7, characterized in that: The information feature analysis model satisfies the following relationship: Among them, Z represents the transaction feature information after analysis, t1 represents the first feature information in the feature label model, σ1 represents the weight corresponding to the first feature, and H 1* represents the first feature information contained in the delivery matrix of the transaction information, t2 represents the second feature information in the feature label model, σ1 represents the weight corresponding to the second feature, and H 2* The second characteristic information contained in the delivery matrix representing the transaction information, t m represents the mth feature information in the feature label model, σ m represents the weight corresponding to the mth feature, H m* The mth feature information contained in the delivery matrix representing the transaction information.

9. The transaction risk control method according to claim 1, characterized in that: The analyzing the platform transaction status based on the transaction feature information and the platform transaction information to ensure risk control and safe operation of the online transaction platform includes: Setting online transaction scheduling indicators based on historical transaction information, wherein the online transaction scheduling indicators include the association status, timing status and feature tag status of online transactions; Analyze the platform transaction status by combining the transaction feature information, the platform transaction information and the online transaction scheduling index, and obtain the operation status of the online transaction platform; The online trading platform is adjusted based on the operating conditions of the online trading platform to achieve risk control and safe operation of the online trading platform.

10. A transaction risk control system, characterized in that: The system includes a data collection module, a data transmission and collation module, a transaction data analysis module and an intelligent trading module. The data collection module, the data transmission and collation module, the transaction data analysis module and the intelligent trading module are interconnected to execute the transaction risk control method as described in any one of claims 1 to 9.

Images