Financial data identification monitoring system and method based on big data

Abstract

The application discloses a financial data recognition monitoring system and method based on big data, which collects multiple user financial transaction data, including transaction amount, transaction time, transaction location, user login information and transaction frequency, and introduces a data processing and analysis algorithm based on artificial intelligence and big data technology in the backend to perform semantic coordination and correlation analysis on the multiple user financial transaction data, so as to capture semantic features related to user financial transaction behavior, and perform abnormal detection on user transaction behavior, so as to identify potential transaction fraud behavior. In this way, the abnormal detection of the user transaction behavior can be realized, so that the abnormal mode of the transaction behavior can be found in time, and the financial fraud behavior can be prevented, and the safety of the user fund is protected.

Description

Technical Field

[0001] This application relates to the field of intelligent monitoring, and more specifically, to a financial data identification and monitoring system and method based on big data. Background Technology

[0002] Financial data refers to various data related to financial transactions, markets, and investments. This data is crucial for financial institutions and regulatory authorities, and can be used for risk management, fraud detection, market analysis, and other areas. As the scale of financial transactions continues to expand, financial fraud is also on the rise. Therefore, the identification and real-time monitoring of anomalies in financial data is of paramount importance.

[0003] However, traditional financial data identification and monitoring systems typically rely on pre-set rules and thresholds to identify abnormal user transaction behavior. This approach is often ill-suited to the complex and ever-changing financial environment, making it prone to missed or false alarms. In other words, traditional systems often use simple pattern recognition techniques, which struggle to handle large-scale, high-dimensional financial data. This limits the accuracy of real-time monitoring and abnormal pattern identification, hindering the comprehensive analysis of fraudulent activities and potential security issues within the financial data, thus posing a threat to user fund security.

[0004] Therefore, an optimized financial data identification and monitoring system is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a financial data identification and monitoring system and method based on big data. It collects multiple user financial transaction data, including transaction amount, transaction time, transaction location, user login information, and transaction frequency. In the backend, it introduces data processing and analysis algorithms based on artificial intelligence and big data technologies to perform semantic collaboration and correlation analysis on these multiple user financial transaction data. This captures semantic features related to user financial transaction behavior and uses them to detect anomalies in user transaction behavior, thereby identifying potential transaction fraud. In this way, it can achieve anomaly detection of user transaction behavior, promptly identify abnormal patterns in transaction behavior, prevent financial fraud, and protect user funds.

[0006] According to one aspect of this application, a big data-based financial data identification and monitoring system is provided, comprising: a financial transaction data acquisition module for acquiring a set of user financial transaction data, wherein the user financial transaction data includes transaction amount, transaction time, transaction location, user login information, and transaction frequency; a financial transaction data embedding and encoding module for performing vector embedding and encoding on each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding and encoding vectors; a single-node financial transaction behavior feature extraction module for extracting single-node financial transaction behavior features from each user financial transaction embedding and encoding vector in the set of user financial transaction embedding and encoding vectors to obtain a set of user financial transaction behavior semantic encoding vectors; a node financial transaction behavior semantic fusion module for inputting the set of user financial transaction behavior semantic encoding vectors into a node message propagation fusion network that fuses node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector; and a financial fraud behavior detection module for determining whether financial fraud behavior exists based on the user financial transaction behavior node transmission aggregate semantic representation vector.

[0007] According to another aspect of this application, a big data-based financial data identification and monitoring method is provided, comprising: acquiring a set of user financial transaction data, wherein the user financial transaction data includes transaction amount, transaction time, transaction location, user login information, and transaction frequency; performing vector embedding encoding on each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding encoding vectors; extracting single-node financial transaction behavior features from each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors to obtain a set of user financial transaction behavior semantic encoding vectors; inputting the set of user financial transaction behavior semantic encoding vectors into a node message propagation fusion network that integrates node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector; and determining whether financial fraud exists based on the user financial transaction behavior node transmission aggregate semantic representation vector.

[0008] Compared with existing technologies, this application provides a big data-based financial data identification and monitoring system and method. It collects multiple user financial transaction data, including transaction amount, transaction time, transaction location, user login information, and transaction frequency. Then, it introduces data processing and analysis algorithms based on artificial intelligence and big data technologies in the backend to perform semantic collaboration and correlation analysis on these multiple user financial transaction data. This captures semantic features related to user financial transaction behavior and uses them to detect anomalies in user transaction behavior, thereby identifying potential transaction fraud. In this way, it can achieve anomaly detection of user transaction behavior, promptly identify abnormal patterns in transaction behavior, prevent financial fraud, and protect user funds. Attached Figure Description

[0009] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 This is a block diagram of a big data-based financial data identification and monitoring system according to an embodiment of this application.

[0011] Figure 2 This is a system architecture diagram of a big data-based financial data identification and monitoring system according to an embodiment of this application.

[0012] Figure 3 This is a block diagram of the semantic fusion module for node financial transaction behavior in a big data-based financial data identification and monitoring system according to an embodiment of this application.

[0013] Figure 4 This is a flowchart of a big data-based financial data identification and monitoring method according to an embodiment of this application. Detailed Implementation

[0014] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0015] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0016] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0017] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0018] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0019] With the rapid development of financial technology, the financial industry has accumulated a large amount of transaction data and user behavior data. The scale and complexity of this data are constantly growing, providing a rich source of information for financial data identification and monitoring systems.

[0020] The technical solution of this application proposes a financial data identification and monitoring system based on big data. Figure 1 This is a block diagram of a big data-based financial data identification and monitoring system according to an embodiment of this application. Figure 2 This is a system architecture diagram of a big data-based financial data identification and monitoring system according to an embodiment of this application. Figure 1 and Figure 2As shown, the big data-based financial data identification and monitoring system 300 according to an embodiment of this application includes: a financial transaction data acquisition module 310, used to acquire a set of user financial transaction data, wherein the user financial transaction data includes transaction amount, transaction time, transaction location, user login information, and transaction frequency; a financial transaction data embedding and encoding module 320, used to perform vector embedding and encoding on each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding and encoding vectors; a single-node financial transaction behavior feature extraction module 330, used to extract single-node financial transaction behavior features from each user financial transaction embedding and encoding vector in the set of user financial transaction embedding and encoding vectors to obtain a set of user financial transaction behavior semantic encoding vectors; a node financial transaction behavior semantic fusion module 340, used to input the set of user financial transaction behavior semantic encoding vectors into a node message propagation fusion network that fuses node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector; and a financial fraud behavior detection module 350, used to determine whether financial fraud behavior exists based on the user financial transaction behavior node transmission aggregate semantic representation vector.

[0021] Specifically, the financial transaction data acquisition module 310 is used to acquire a collection of user financial transaction data, which includes transaction amount, transaction time, transaction location, user login information, and transaction frequency. In a specific example, transaction amount is the amount of money involved in each user transaction. This information is crucial for understanding the user's consumption habits, financial situation, and the importance of the transactions. By analyzing transaction amount, the user's consumption level, preferences, and transaction patterns can be inferred. Transaction time records the specific time when the user conducts a transaction. Transaction time can reveal the user's transaction habits, activity patterns, and possible abnormal behaviors. Analyzing transaction time can help predict the user's future transaction behavior, optimize service hours, and detect potential fraudulent activities. Transaction location refers to the geographical location information of the user's transaction. This information is important for understanding the user's activity range, travel habits, and possible risky behaviors. By analyzing transaction location, more personalized services can be provided, frequently used locations of users can be identified, and the risks of cross-regional transactions can be detected. User login information includes the user's account login time, device information, login location, etc. This information can be used to verify user identity, monitor account security, and identify abnormal login behavior. Transaction frequency refers to the number of times a user conducts transactions within a certain period. Transaction frequency reflects a user's activity level, trading habits, and potential changes. Analyzing transaction frequency allows for understanding user trading behavior patterns and identifying abnormal trading behavior. Therefore, the technical solution in this application collects multiple user financial transaction data, including transaction amount, transaction time, transaction location, user login information, and transaction frequency. On the backend, it introduces data processing and analysis algorithms based on artificial intelligence and big data technologies to perform semantic collaboration and correlation analysis on these multiple user financial transaction data. This captures semantic features related to user financial transaction behavior and uses them to detect anomalies in user trading behavior, thereby identifying potential transaction fraud. This enables the detection of anomalies in user trading behavior, allowing for the timely discovery of abnormal trading patterns, preventing financial fraud, and protecting user funds.

[0022] Specifically, the financial transaction data embedding and encoding module 320 is used to perform vector embedding and encoding on each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding and encoding vectors. Considering that each financial transaction data includes information of various formats and types, such as transaction amount, transaction time, and transaction location, and that these different types of transaction information have interrelationships to determine whether a user's transaction behavior constitutes financial fraud, in order to capture the semantic features of each financial transaction information in the financial transaction data, thereby providing a basis for subsequent semantic association analysis of financial transaction data and anomaly detection of user financial transaction behavior, the technical solution of this application requires vector embedding and encoding on each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding and encoding vectors. Semantic embedding and encoding can convert the format of each user financial transaction data, transforming these heterogeneous data into a unified numerical vector form, facilitating subsequent processing and analysis. Furthermore, embedding encoding can help the system extract key semantic features of each financial transaction in each financial transaction data. These features are crucial for understanding transaction behavior and identifying fraud patterns, providing support for subsequent anomaly detection and fraud identification.

[0023] Specifically, the single-node financial transaction behavior feature extraction module 330 is used to extract user financial transaction behavior features from each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors to obtain a set of user financial transaction behavior semantic encoding vectors. In a specific example of this application, each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors is passed through a single-node financial transaction behavior feature extractor based on a fully connected layer to obtain the set of user financial transaction behavior semantic encoding vectors. Considering that although each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors contains embedded semantic information about various information in the user financial transaction data, it cannot understand the semantic relationships and hidden features between various financial transaction information. Such relationship features can better reflect the user's transaction behavior and are beneficial for anomaly detection in financial transaction behavior. Based on this, in the technical solution of this application, each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors is further passed through a single-node financial transaction behavior feature extractor based on a fully connected layer to obtain the set of user financial transaction behavior semantic encoding vectors. Through the processing of the single-node financial transaction behavior feature extractor based on the fully connected layer, the transaction information of each user's financial transaction data can be embedded into the encoded semantics for correlation analysis and feature extraction, thereby capturing the semantic features of transaction behavior in each user's financial transaction data. This means that they can better represent the inherent behavioral meaning and contextual information in each user's financial transaction data, which helps to identify more complex transaction behavior patterns and provides support for identifying and preventing users from being defrauded.

[0024] Specifically, the node financial transaction behavior semantic fusion module 340 is used to input the set of user financial transaction behavior semantic encoding vectors into a node message propagation fusion network that fuses node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector. Considering that each user financial transaction behavior semantic encoding vector in the set contains semantic features related to each user's financial transaction behavior, and that these financial transaction behavior semantics have hidden correlations, this is crucial for a more comprehensive understanding of the user's overall financial behavior and for identifying and detecting abnormal user transaction behavior. Based on this, in the technical solution of this application, the set of user financial transaction behavior semantic encoding vectors is further input into a node message propagation fusion network that fuses node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector. Through the processing of the node message propagation fusion network that fuses node importance, information transmission aggregation analysis can be performed on the semantic encoding features of each user's financial transaction behavior.

[0025] Specifically, when fusing the semantic features of financial transaction behavior at each node, the node message propagation fusion network can dynamically update the node representation to reflect the latest transaction behavior. Furthermore, it can assign different weights based on the importance of the transaction behaviors to adapt to different users' financial transaction data. This allows for better modeling of the relationships between users in each financial transaction, capturing the potential relationships and pattern features between nodes in each financial transaction, thus obtaining a more global semantic representation of financial transaction behavior. This helps the system to more comprehensively understand the user's overall financial behavior. Specifically, in a specific example of this application, such as... Figure 3 As shown, the node financial transaction behavior semantic fusion module 340 includes: a user financial transaction behavior pre-information fusion unit 341, used to fuse the pre-information of the user financial transaction behavior semantic encoding vector set. The semantic information of the semantic encoding vector of a user's financial transaction behavior is fused to obtain the semantic fusion feature vector of the user's financial transaction behavior pre-information; the semantic weighting optimization unit 342 is used to optimize the semantic information of the first user's financial transaction behavior semantic encoding vector. The semantic weighting optimization of the semantic encoding vector of the user's financial transaction behavior is performed on the first user financial transaction behavior to obtain a weighted optimized semantic encoding vector of the user financial transaction behavior; the user financial transaction behavior importance reasoning perception unit 343 is used to perform position-wise summation of the semantic fusion feature vector of the user financial transaction behavior pre-information and the weighted optimized semantic encoding vector of the user financial transaction behavior and then process it through a multilayer perceptron to obtain the first user financial transaction behavior semantic encoding vector. The semantic feature vector for the importance inference of user financial transaction behavior corresponding to each semantic encoding vector of user financial transaction behavior; the user financial transaction behavior node transmission aggregation unit 344 is used to calculate the positional summation between the semantic feature vectors for the importance inference of user financial transaction behavior corresponding to each semantic encoding vector of user financial transaction behavior in the set of semantic encoding vectors of user financial transaction behavior to obtain the aggregated semantic representation vector for the transmission of user financial transaction behavior nodes.

[0026] Specifically, the user financial transaction behavior pre-information fusion unit 341 is used to integrate the pre-information from the set of user financial transaction behavior semantic encoding vectors. The semantic information of the semantic encoding vectors of individual user financial transaction behaviors is fused to obtain the semantic fusion feature vector of the preceding information of user financial transaction behavior. In a specific example of this application, the preceding information of the semantic encoding vectors of individual user financial transaction behaviors is calculated. The semantic encoding vectors of each user's financial transaction behavior are summed positionally to obtain the semantic fusion feature vector of the user's financial transaction behavior pre-information.

[0027] Specifically, the semantic weight weighting optimization unit 342 is used to optimize the semantic weights of the first... The semantic weighting optimization of the semantic encoding vectors of user financial transaction behavior is performed on the first user financial transaction behavior to obtain a weighted optimized semantic encoding vector of user financial transaction behavior. In a specific example of this application, the sum of trainable preset hyperparameters and one is used as the weighting coefficient for the first user financial transaction behavior semantic encoding vector. The feature values ​​at each position in the semantic encoding vector of the user's financial transaction behavior are weighted to obtain the weighted optimized semantic encoding vector of the user's financial transaction behavior.

[0028] Specifically, the user financial transaction behavior importance inference perception unit 343 is used to sum the semantic fusion feature vector of the user financial transaction behavior prior information and the weighted optimized user financial transaction behavior semantic encoding vector by position, and then process them through a multilayer perceptron to obtain the first... The semantic feature vectors corresponding to the semantic encoding vectors of user financial transaction behavior are the semantic feature vectors for inferring the importance of user financial transaction behavior. Among them, the Multilayer Perceptron (MLP) is a basic form of artificial neural network. It consists of multiple neural network layers, each containing multiple neurons (also called nodes).

[0029] Accordingly, in one possible implementation, the semantic fusion feature vector of the user's financial transaction behavior pre-information and the weighted optimized semantic encoding vector of the user's financial transaction behavior can be summed positionally and then processed by a multilayer perceptron to obtain the first... The semantic feature vector for inferring the importance of user financial transaction behavior is obtained by, for example: inputting the semantic fusion feature vector of the user financial transaction behavior's prior information and the weighted optimized semantic encoding vector of the user financial transaction behavior; performing a positional summation operation on the semantic fusion feature vector of the user financial transaction behavior's prior information and the weighted optimized semantic encoding vector of the user financial transaction behavior to obtain a new feature vector; inputting the obtained feature vector into a multilayer perceptron for processing to infer the importance of user financial transaction behavior; using a labeled training dataset, training the multilayer perceptron through the backpropagation algorithm, adjusting the weights and biases to minimize the loss function; to obtain the semantic feature vector for inferring the importance of user financial transaction behavior.

[0030] Specifically, the user financial transaction behavior node transmission aggregation unit 344 is used to calculate the positional summation of the user financial transaction behavior importance inference semantic feature vectors corresponding to each user financial transaction behavior semantic encoding vector in the set of user financial transaction behavior semantic encoding vectors to obtain the user financial transaction behavior node transmission aggregated semantic representation vector. It should be understood that calculating the positional summation of the user financial transaction behavior importance inference semantic feature vectors corresponding to each user financial transaction behavior semantic encoding vector in the set of user financial transaction behavior semantic encoding vectors can provide a more comprehensive, stable, and richer feature representation, thereby helping to better understand and utilize the information in the financial transaction behavior data.

[0031] In summary, in the above embodiments, inputting the set of semantic encoding vectors of user financial transaction behavior into the node message propagation fusion network that integrates node importance to obtain the aggregate semantic representation vector of user financial transaction behavior node transmission includes: inputting the set of semantic encoding vectors of user financial transaction behavior into the node message propagation fusion network that integrates node importance and processing it with the following node message propagation formula to obtain the aggregate semantic representation vector of user financial transaction behavior node transmission; wherein, the information transmission inference formula is: ;in, It is the first in the set of semantic encoding vectors of the user's financial transaction behavior. Semantic encoding vector of a user's financial transaction behavior It is the first in the set of semantic encoding vectors of the user's financial transaction behavior. Semantic encoding vector of a user's financial transaction behavior It is a trainable preset hyperparameter. This represents a multilayer perceptron. The number of feature vectors in the set of semantic encoding vectors of the user's financial transaction behavior. The aggregated semantic representation vector is passed to the user's financial transaction behavior node.

[0032] It is worth mentioning that, in other specific examples of this application, the set of semantic encoding vectors of user financial transaction behavior can also be input into a node message propagation fusion network that integrates node importance to obtain the aggregated semantic representation vector of user financial transaction behavior node transmission. For example: input the set of semantic encoding vectors of user financial transaction behavior; design a network structure that includes a mechanism for updating node representations and propagating messages to propagate and integrate node information in the network; propagate the information of each node to its neighboring nodes according to the connection relationship and weight between nodes; combine the propagated information and the node's own information and perform weighted fusion according to node importance; update the representation of each node according to the propagation and fusion results; for all nodes, further aggregation operations can be performed based on the updated node representations to obtain the aggregated semantic representation vector of user financial transaction behavior node transmission.

[0033] Specifically, the financial fraud detection module 350 is used to determine whether financial fraud exists based on the aggregated semantic representation vector transmitted by the user's financial transaction behavior node. In a specific example of this application, the aggregated semantic representation vector transmitted by the user's financial transaction behavior node is passed through a classifier-based identification monitor to obtain an identification monitoring result, which indicates whether financial fraud exists. That is, the aggregated semantic features transmitted by the user's financial transaction behavior node are used for classification processing to detect abnormal user transaction behavior and thus identify potential transaction fraud. In this way, abnormal detection of user transaction behavior can be achieved, so as to promptly discover abnormal patterns of transaction behavior and prevent financial fraud, thereby protecting the user's funds. Specifically, the aggregated semantic representation vector transmitted by the user's financial transaction behavior node is passed through a classifier-based identification monitor to obtain an identification monitoring result, which indicates whether financial fraud exists. This includes: using multiple fully connected layers of the classifier to fully connect and encode the aggregated semantic representation vector transmitted by the user's financial transaction behavior node to obtain an encoded classification feature vector; and passing the encoded classification feature vector through the classifier's Softmax classification function to obtain the identification monitoring result.

[0034] In other words, in the technical solution of this application, the classifier's labels include "existence of financial fraud" (first label) and "absence of financial fraud" (second label). The classifier uses a soft-maximum function to determine which category label the user's financial transaction behavior node's aggregated semantic representation vector belongs to. It is worth noting that the first label p1 and the second label p2 here do not contain artificially defined concepts. In fact, during the training process, the computer model does not have the concept of "whether financial fraud exists." It simply has two category labels and outputs the probability of the feature under these two category labels, i.e., the sum of p1 and p2 is one. Therefore, the classification result of whether financial fraud exists is actually transformed into a binary probability distribution that conforms to natural laws through the category labels. Essentially, it uses the physical meaning of the natural probability distribution of the labels, rather than the linguistic textual meaning of "whether financial fraud exists."

[0035] It's worth noting that a classifier is a machine learning model or algorithm used to categorize input data into different classes or labels. Classifiers are a part of supervised learning; they perform classification tasks by learning a mapping from input data to output classes.

[0036] In the technical solution of this application, each user financial transaction behavior semantic encoding vector in the set of user financial transaction behavior semantic encoding vectors is used to represent the implicit financial transaction behavior semantic encoding features in each user financial transaction data. When the set of user financial transaction behavior semantic encoding vectors is input into the node message propagation fusion network that integrates node importance, the network aggregates the implicit financial transaction behavior semantic encoding features in each user financial transaction data along the time dimension to capture the temporal implicit correlation features of user financial transaction behavior. This allows for the differentiated transmission and aggregation of information based on the node importance of the implicit financial transaction behavior semantic encoding features in each user financial transaction data, making the user financial transaction behavior node transmission aggregated semantic representation vector have significant feature distribution in the time dimension. However, this also causes the feature distribution of the user financial transaction behavior node transmission aggregated semantic representation vector to have local overflow characteristics, leading to regression pattern shift. This results in overflow of the monitoring results obtained by the classifier-based recognition monitor through the user financial transaction behavior node transmission aggregated semantic representation vector, affecting its accuracy.

[0037] Based on this, obtaining the identification and monitoring results by passing the aggregated semantic representation vector of the user's financial transaction behavior nodes through a classifier-based identification and monitoring system specifically includes: passing the aggregated semantic representation vector of the user's financial transaction behavior nodes through a probabilistic activation function to obtain a probabilistic aggregated semantic representation vector of the user's financial transaction behavior nodes; determining the abnormal probability value of the representation of financial fraud behavior obtained by passing the aggregated semantic representation vector of the user's financial transaction behavior nodes through the classifier-based identification and monitoring system; performing a dot subtraction between the unit feature vector and the probabilistic aggregated semantic representation vector of the user's financial transaction behavior nodes, using this as the base to calculate a power function with the abnormal probability value as the exponent, and further performing a dot product with the probabilistic aggregated semantic representation vector of the user's financial transaction behavior nodes to obtain a probabilistic aggregated semantic representation vector of the user's financial transaction behavior nodes. The semantic convergence constraint feature vector is obtained by: using the absolute value of each feature value of the probabilistic user financial transaction behavior node transitive aggregate semantic representation vector as the base, calculating a power function with the reciprocal of the anomaly probability value as the exponent, and summing all feature values ​​of the probabilistic user financial transaction behavior node transitive aggregate semantic representation vector to obtain the user financial transaction behavior node transitive aggregate semantic convergence bias feature value; summing the product of the probabilistic user financial transaction behavior node transitive aggregate semantic convergence constraint feature vector and the user financial transaction behavior node transitive aggregate semantic convergence bias feature value with the weights as hyperparameters to obtain the optimized user financial transaction behavior node transitive aggregate semantic representation vector; and passing the optimized user financial transaction behavior node transitive aggregate semantic representation vector through the classifier-based recognition monitor to obtain the recognition monitoring result.

[0038] Let the probabilized user financial transaction behavior node transmit aggregate semantic representation vector be denoted as . The abnormal probability value is The optimized user financial transaction behavior node transmits the aggregated semantic representation vector. Represented as: ;in, It is the probabilistic user financial transaction behavior node that transmits the aggregated semantic representation vector. It is a unit eigenvector where all eigenvalues ​​are one. These are the weights of the hyperparameters. It is the first of the probabilistic user financial transaction behavior nodes that transmits the aggregated semantic representation vector. 1 eigenvalue, The anomaly probability value is... It is a power function that calculates the feature values ​​at each position in the feature vector using the anomaly probability value as an exponent. This indicates dot product by position. This indicates the difference based on position. This indicates addition by position. It is the optimized user financial transaction behavior node transmitted aggregate semantic representation vector.

[0039] In other words, for the aggregated semantic representation vector of the user's financial transaction behavior node, based on the convergence constraint of the Cauchy-Hadamard form power series probability distribution, the feature set of the aggregated semantic representation vector of the user's financial transaction behavior node is represented by a power series based on the convergence probability. A bounded convergence bias is added, using each feature value of the aggregated semantic representation vector of the user's financial transaction behavior node as the convergence radius. This avoids invalid overflow of the classification results caused by the deviation of the distribution pattern of the aggregated semantic representation vector of the user's financial transaction behavior node during the regression process of the classifier-based identification monitor due to the gradient propagation of model parameters based on the convergence probability, under a unified probability convergence constraint standard. This improves the accuracy of the identification and monitoring results obtained by the classifier-based identification monitor. In this way, abnormal detection of user transaction behavior can be performed more accurately, enabling timely identification and prevention of potential transaction fraud and protecting user funds.

[0040] As described above, the big data-based financial data identification and monitoring system 300 according to embodiments of this application can be implemented in various wireless terminals, such as servers with big data-based financial data identification and monitoring algorithms. In one possible implementation, the big data-based financial data identification and monitoring system 300 according to embodiments of this application can be integrated into the wireless terminal as a software module and / or hardware module. For example, the big data-based financial data identification and monitoring system 300 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the big data-based financial data identification and monitoring system 300 can also be one of many hardware modules of the wireless terminal.

[0041] Alternatively, in another example, the big data-based financial data identification and monitoring system 300 and the wireless terminal can also be separate devices, and the big data-based financial data identification and monitoring system 300 can be connected to the wireless terminal via wired and / or wireless networks, and transmit interactive information in accordance with the agreed data format.

[0042] Furthermore, a method for financial data identification and monitoring based on big data is also provided.

[0043] Figure 4 This is a flowchart of a big data-based financial data identification and monitoring method according to an embodiment of this application. Figure 4As shown, the big data-based financial data identification and monitoring method according to an embodiment of this application includes: S1, acquiring a set of user financial transaction data, wherein the user financial transaction data includes transaction amount, transaction time, transaction location, user login information, and transaction frequency; S2, performing vector embedding encoding on each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding encoding vectors; S3, extracting single-node financial transaction behavior features from each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors to obtain a set of user financial transaction behavior semantic encoding vectors; S4, inputting the set of user financial transaction behavior semantic encoding vectors into a node message propagation fusion network that integrates node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector; S5, determining whether financial fraud exists based on the user financial transaction behavior node transmission aggregate semantic representation vector.

[0044] In summary, the big data-based financial data identification and monitoring method according to the embodiments of this application is explained. It collects multiple user financial transaction data, including transaction amount, transaction time, transaction location, user login information, and transaction frequency. Then, it introduces data processing and analysis algorithms based on artificial intelligence and big data technologies in the backend to perform semantic collaboration and correlation analysis on these multiple user financial transaction data. This captures semantic features related to user financial transaction behavior and uses them to detect anomalies in user transaction behavior, thereby identifying potential transaction fraud. In this way, it can achieve anomaly detection of user transaction behavior, promptly identify abnormal patterns in transaction behavior, prevent financial fraud, and protect user funds.

[0045] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A financial data identification and monitoring system based on big data, characterized in that, include: The system includes a financial transaction data acquisition module for acquiring a set of user financial transaction data, including transaction amount, transaction time, transaction location, user login information, and transaction frequency; a financial transaction data embedding and encoding module for embedding and encoding each user financial transaction data in the set of user financial transaction data to obtain a set of user financial transaction embedding and encoding vectors; a single-node financial transaction behavior feature extraction module for extracting single-node financial transaction behavior features from each user financial transaction embedding and encoding vector in the set of user financial transaction embedding and encoding vectors to obtain a set of user financial transaction behavior semantic encoding vectors; a node financial transaction behavior semantic fusion module for inputting the set of user financial transaction behavior semantic encoding vectors into a node message propagation fusion network that fuses node importance to obtain a user financial transaction behavior node transmission aggregate semantic representation vector; and a financial fraud behavior detection module for determining whether financial fraud behavior exists based on the user financial transaction behavior node transmission aggregate semantic representation vector. The single-node financial transaction behavior feature extraction module is used to: pass each user financial transaction embedding encoding vector in the set of user financial transaction embedding encoding vectors through a single-node financial transaction behavior feature extractor based on a fully connected layer to obtain the set of user financial transaction behavior semantic encoding vectors. The node financial transaction behavior semantic fusion module includes: a user financial transaction behavior pre-information fusion unit, used to fuse the first... The first set of semantic encoding vectors of individual user financial transaction behavior The semantic information of the semantic encoding vector of a user's financial transaction behavior is fused to obtain the semantic fusion feature vector of the user's financial transaction behavior pre-information; the semantic weighting optimization unit is used to optimize the semantic information of the first user's financial transaction behavior semantic encoding vector. The semantic weighting optimization of the semantic encoding vectors of the user's financial transaction behavior is performed to obtain a weighted optimized semantic encoding vector of the user's financial transaction behavior; the user financial transaction behavior importance inference perception unit is used to perform position-wise summation of the semantic fusion feature vector of the user's financial transaction behavior pre-information and the weighted optimized semantic encoding vector of the user's financial transaction behavior, and then process it through a multilayer perceptron to obtain the first... The semantic feature vector for the importance inference of user financial transaction behavior corresponding to each semantic encoding vector of user financial transaction behavior; the user financial transaction behavior node transmission aggregation unit is used to calculate the positional summation between the semantic feature vectors for the importance inference of user financial transaction behavior corresponding to each semantic encoding vector of user financial transaction behavior in the set of semantic encoding vectors of user financial transaction behavior to obtain the aggregated semantic representation vector for the transmission of user financial transaction behavior nodes.

2. The financial data identification and monitoring system based on big data according to claim 1, characterized in that, The user financial transaction behavior pre-information fusion unit is used to: calculate the first... The first set of semantic encoding vectors of individual user financial transaction behavior The semantic encoding vectors of each user's financial transaction behavior are summed positionally to obtain the semantic fusion feature vector of the user's financial transaction behavior pre-information.

3. The financial data identification and monitoring system based on big data according to claim 2, characterized in that, The semantic weighting optimization unit is used to: use the sum of trainable preset hyperparameters and one as weighting coefficients to optimize the semantic weighting of the first... The feature values ​​at each position in the semantic encoding vector of the user's financial transaction behavior are weighted to obtain the weighted optimized semantic encoding vector of the user's financial transaction behavior.

4. The financial data identification and monitoring system based on big data according to claim 3, characterized in that, The financial fraud detection module is used to: pass the user's financial transaction behavior node to an aggregated semantic representation vector and pass it through a classifier-based identification monitor to obtain an identification monitoring result, wherein the identification monitoring result is used to indicate whether financial fraud behavior exists.

5. The financial data identification and monitoring system based on big data according to claim 4, characterized in that, The financial fraud detection module includes: a fully connected encoding unit, used to perform fully connected encoding on the aggregate semantic representation vector of the user's financial transaction behavior node using multiple fully connected layers of the classifier to obtain an encoded classification feature vector; and a classification result generation unit, used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the identification and monitoring result.

6. A financial data identification and monitoring method based on big data, using the financial data identification and monitoring system based on big data as described in claim 1, characterized in that, include: A set of user financial transaction data is obtained, wherein the user financial transaction data includes transaction amount, transaction time, transaction location, user login information and transaction frequency; each user financial transaction data in the set of user financial transaction data is vector-embedded and encoded to obtain a set of user financial transaction embedding and encoded vectors; each user financial transaction embedding and encoded vector in the set of user financial transaction embedding and encoded vectors is subjected to single-node financial transaction behavior feature extraction to obtain a set of user financial transaction behavior semantic encoding vectors; The set of semantic encoding vectors of the user's financial transaction behavior is input into the node message propagation fusion network of the fusion node importance to obtain the aggregate semantic representation vector of the user's financial transaction behavior node transmission; based on the aggregate semantic representation vector of the user's financial transaction behavior node transmission, it is determined whether there is financial fraud.

Images