Data processing method and device, equipment, storage medium and product
By constructing heterogeneous graphs and graph attention analysis, combined with multi-source data and risk assessment models, the problem of insufficient accuracy in risk identification during marketing activities is solved, enabling efficient risk identification and real-time early warning for marketing events.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-25
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies fail to effectively consider the constraints between indicators in marketing campaign risk identification, resulting in insufficient accuracy in risk identification.
Heterogeneous graphs are constructed using multi-source heterogeneous data, and risk indicators and their constraints are determined through graph attention analysis. Risk identification is then performed by combining risk assessment models and external environmental data.
It improves the accuracy and precision of marketing event risk identification, can identify abnormal risk indicators and provide real-time early warnings, and enhances the effectiveness of risk management.
Smart Images
Figure CN121786689A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of financial technology and risk management technology, and in particular relates to a data processing method, apparatus, device, storage medium and product. Background Technology
[0002] With the rapid development of financial markets and the increasing diversification of customer needs, marketing activities of financial institutions play a vital role in attracting customers and enhancing brand influence. While marketing activities bring business growth, they also bring a series of management risks, such as the malicious exploitation of rule loopholes, deviation from activity objectives, resource misallocation, and compliance risks. Therefore, identifying the risks in marketing activities is of great importance.
[0003] In the process of risk identification for marketing campaigns, it is necessary to identify risk indicators of data anomalies. However, in related technologies, the risk identification process for marketing campaigns typically involves analyzing individual indicators separately without considering the constraints between indicators. This approach fails to identify risk indicators, thus failing to identify abnormal risks in marketing campaigns and reducing the accuracy of risk identification. Summary of the Invention
[0004] This application provides a data processing method, apparatus, device, storage medium, and product that can improve the accuracy of marketing event risk identification.
[0005] In a first aspect, embodiments of this application provide a data processing method, which includes: collecting multi-source heterogeneous data generated during a marketing event, wherein the multi-source heterogeneous data includes at least user behavior data of participating users in the marketing event, marketing result data corresponding to the marketing event, and transaction data corresponding to the marketing event; constructing a heterogeneous graph corresponding to the multi-source heterogeneous data; performing graph attention analysis on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraint relationships between the multiple risk indicators; extracting indicator data corresponding to each risk indicator from the multi-source heterogeneous data; and determining abnormal risk indicators from the multiple risk indicators based on the constraint relationships between the multiple risk indicators and the indicator data.
[0006] In some possible implementations, multi-source heterogeneous data generated during the marketing event is collected, including: collecting user behavior data of participating users through tracking tools set in the marketing platform, wherein the tracking tools are set at the location of the functional scripts that implement the marketing event in the marketing platform; collecting marketing result data corresponding to the marketing event from the data warehouse through the data warehouse exchange interface, wherein the data warehouse is used to store the result data of the marketing event; and collecting transaction data corresponding to the marketing event from the transaction platform through the transaction interface.
[0007] In some possible implementations, a heterogeneous graph corresponding to the multi-source heterogeneous data is constructed, including: performing data fusion on the multi-source heterogeneous data to obtain fused data; extracting entities corresponding to marketing events and the relationships between entities from the fused data; and constructing a heterogeneous graph corresponding to the marketing events based on the entities and the relationships between entities. The heterogeneous graph includes nodes and edges, where nodes are used to represent entities in the marketing event, and edges are used to represent the relationships between entities, including event participation relationships, transaction relationships, and attribution relationships.
[0008] In some possible implementations, multi-source heterogeneous data is fused to obtain fused data, including: parsing and filtering user behavior data through a data processing engine to obtain first data; converting marketing result data and transaction data to obtain second data; and fusing the first data and second data with the user identifiers of participating users and a preset time window to obtain fused data.
[0009] In some possible implementations, multiple risk indicators include first-type and second-type indicators. Graph attention analysis is performed on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraints between them. This includes: responding to an indicator selection instruction, selecting first-type and second-type indicators from an indicator library, where the first-type indicators are key indicators necessary for risk assessment of all marketing events, and the second-type indicators are indicators matched to the risk assessment needs corresponding to the marketing events; obtaining graph data corresponding to the first-type and second-type indicators from the heterogeneous graph; inputting the graph data into a graph attention network to obtain the constraints between the multiple risk indicators output by the graph attention network, where the graph attention network is used to determine the feature representation of each node in the heterogeneous graph under the influence of its neighboring nodes based on the graph data, and to calculate the importance of different risk indicators to risk propagation through an attention mechanism to obtain the constraints.
[0010] In some possible implementations, abnormal risk indicators are determined from multiple risk indicators based on the constraints between multiple risk indicators and the indicator data. This includes: obtaining the initial risk threshold corresponding to the marketing event; inputting the indicator data corresponding to multiple risk indicators into the risk assessment model to obtain the initial risk score output by the risk assessment model; and determining the abnormal risk indicators based on the initial risk score and the initial risk threshold.
[0011] In some possible implementations, abnormal risk indicators are determined based on the initial risk score and the initial risk threshold, including: obtaining external environmental data corresponding to the marketing event; determining a first adjustment coefficient for adjusting the initial risk score and a second adjustment coefficient for adjusting the initial risk threshold based on the relationship between the environmental data and the adjustment coefficient; adjusting the initial risk score and the initial risk threshold using the first adjustment coefficient and the second adjustment coefficient respectively to obtain the target risk score and the target risk threshold; and determining the abnormal risk indicators based on the relative value between the target risk score and the target risk threshold.
[0012] In some possible implementations, after determining the abnormal risk indicators based on the relative value between the target risk score and the target risk threshold, attribution analysis is performed on the indicator data to determine the contribution of multiple risk indicators to the target risk score; target risk indicators whose contribution is within a preset range are obtained; and the risk transmission path corresponding to the marketing event is determined based on the target risk indicators.
[0013] In some possible implementations, after determining abnormal risk indicators from multiple risk indicators based on the constraint relationships between multiple risk indicators and the indicator data, the marketing attribute data corresponding to the marketing event is obtained; based on the marketing attribute data, the risk handling strategy corresponding to at least one historical marketing event is obtained from the strategy knowledge base, wherein at least one historical marketing event is a marketing event whose similarity to the marketing attribute data of the marketing event is greater than a similarity threshold; and the risk handling strategy is displayed.
[0014] Secondly, embodiments of this application provide a data processing apparatus, comprising: a data acquisition module for acquiring multi-source heterogeneous data generated during a marketing event, wherein the multi-source heterogeneous data includes at least user behavior data of participating users in the marketing event, marketing result data corresponding to the marketing event, and transaction data corresponding to the marketing event; a graph construction module for constructing a heterogeneous graph corresponding to the multi-source heterogeneous data; an indicator determination module for performing graph attention analysis on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraint relationships between the multiple risk indicators; a data extraction module for extracting indicator data corresponding to each risk indicator from the multi-source heterogeneous data; and a risk identification module for determining abnormal risk indicators from the multiple risk indicators based on the constraint relationships between the multiple risk indicators and the indicator data.
[0015] Thirdly, embodiments of this application provide an electronic device, which includes: a processor and a memory storing computer program instructions; and a data processing method that implements any of the above-mentioned methods when the processor executes the computer program instructions.
[0016] Fourthly, embodiments of this application provide a computer storage medium on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the data processing method described above is implemented.
[0017] Fifthly, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, enable the electronic device to perform any of the above-mentioned data processing methods.
[0018] As described above, this embodiment employs heterogeneous graphs and graph attention mechanisms to analyze multi-source heterogeneous data, determining indicators for identifying marketing event risks, thus enabling risk identification of marketing events. In this embodiment, multi-source heterogeneous data is collected in real-time during the marketing event process, including not only user behavior data but also marketing outcome data and transaction data generated by the marketing event. A multi-source data fusion approach is used for risk identification, improving the accuracy of risk identification. Furthermore, this embodiment utilizes heterogeneous graphs and graph attention analysis to extract features from multi-source heterogeneous data. This approach considers the strong correlation between risk indicators, thereby improving the accuracy of marketing event risk identification. Attached Figure Description
[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a schematic diagram of the structure of a data processing system provided in one embodiment of this application; Figure 2 This is a schematic flowchart of a data processing method provided in one embodiment of this application; Figure 3 This is a schematic diagram of the interface for risk identification provided in one embodiment of this application; Figure 4 This is a schematic diagram of the structure of a data processing apparatus provided in another embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in another embodiment of this application. Detailed Implementation
[0021] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.
[0022] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.
[0023] It should be noted that the acquisition, storage, use, and processing of data in this application embodiment all comply with the relevant provisions of national laws and regulations.
[0024] Furthermore, it should be noted that in the embodiments of this application, certain software, components, models, and other existing solutions in the industry may be mentioned. These should be considered as exemplary, and their purpose is only to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.
[0025] With the rapid development of financial markets and the increasing diversification of customer needs, marketing activities of financial institutions play a vital role in attracting customers and enhancing brand influence. However, these activities also come with potential management risks, making effective assessment and early warning of these risks particularly important.
[0026] Currently, financial institutions' marketing activity risk assessments primarily rely on traditional risk management frameworks, which typically include steps such as risk identification, risk assessment, and risk control. Specifically, this involves identifying the business parameters to be assessed, obtaining a set of assessment indicators for these parameters, inputting this set into a pre-trained risk assessment model, and outputting the risk assessment results for the business parameters. Furthermore, related technologies monitor transaction data and behaviors across various business systems, analyze abnormal and non-compliant events, and conduct risk assessments of institutions, products, positions, personnel, and processes to achieve automated supervision and error handling of key business operations.
[0027] However, these frameworks have significant limitations when dealing with large amounts of activity data, such as the inability to process and analyze data in real time and the lack of in-depth analysis of activity effectiveness. Moreover, the aforementioned methods are mainly based on expert experience and subjective judgment for risk identification and assessment.
[0028] Furthermore, some financial institutions have attempted to use machine learning and artificial intelligence technologies to improve the accuracy of risk assessment. However, these attempts are often limited to single risk types or specific data sources, failing to fully cover the complexity and diversity of financial institutions' marketing activities. Specifically, existing big data risk control technologies face problems such as insufficient data completeness and accuracy, significant subjective factors, and low accuracy. Moreover, data access is difficult, and precise quantitative analysis is lacking. Significant difficulties are encountered in the process of accessing external data, including inconsistent external data quality, difficulty in assessing data utility, and the inability to implement traditional risk control data collection methods for online business.
[0029] In summary, existing risk assessment methods for marketing activities of financial institutions are inadequate in terms of data processing capabilities, analytical depth, and early warning accuracy, and a new technological solution is urgently needed to improve the effectiveness of risk management.
[0030] To address the problems of the prior art, embodiments of the present invention provide a data processing method, apparatus, device, storage medium, and product.
[0031] In this application embodiment, the data processing system can serve as the execution entity of the method provided in this application embodiment, such as... Figure 1As shown, the data processing system includes a data acquisition unit 10, a data preprocessing unit 11, an indicator engineering unit 12, a real-time monitoring and early warning unit 13, and a display unit 14. The data acquisition unit collects user behavior data, marketing result data, and transaction data according to unified indicator dimensions. The data preprocessing unit cleans, transforms, and standardizes the collected data to ensure data quality and provide accurate input. The indicator engineering unit extracts and selects features relevant to risk identification and assessment from the preprocessed data, anchors existing indicators, or expands new ones. Through real-time statistical analysis of massive amounts of data, it identifies key risk indicators. The real-time monitoring and early warning unit monitors the data flow of marketing events in real time, performs risk assessment on the data, and automatically triggers an early warning mechanism based on key risk indicator data. The display unit provides a user-friendly interface, displaying risk identification results and early warning information.
[0032] The data processing method provided in the embodiments of this application will be described below.
[0033] Figure 2 A schematic flowchart of a data processing method according to an embodiment of the present invention is shown. Figure 2 As shown, the method may include the following steps S201 to S205: Step S201: Collect multi-source heterogeneous data generated during the marketing event.
[0034] In step S201, the marketing event can be a marketing activity organized by a financial institution, such as a personal pension appointment gift activity or a monthly phone bill red envelope activity.
[0035] In this embodiment of the application, multi-source heterogeneous data refers to data generated during the marketing event. Multi-source heterogeneous data includes at least user behavior data of participating users in the marketing event (e.g., the number of times users browse the event and the duration of browsing), marketing result data corresponding to the marketing event (e.g., the number of users participating in the marketing event, the number of users who meet the criteria, the number of customers who receive prizes, etc.), and transaction data corresponding to the marketing event (transaction time, transaction amount, etc.).
[0036] In this application embodiment, different data collection interfaces are set up to collect corresponding data for different data sources. Compared with the existing technology of collecting single data and conducting risk analysis of marketing events based on single data, this application embodiment uses multi-source heterogeneous data, which can improve the accuracy of risk analysis.
[0037] Step S202: Construct a heterogeneous graph corresponding to the multi-source heterogeneous data.
[0038] In step S202, a heterogeneous graph is a graph data structure containing various types of nodes and edges. In this embodiment, by constructing a heterogeneous graph corresponding to multi-source heterogeneous data, internal activity data of marketing events, marketing result data, user behavior relationship data, etc., risk features can be extracted and enhanced through the heterogeneous graph.
[0039] Step S203: Perform graph attention analysis on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraint relationships between them.
[0040] In step S203, multiple risk indicators may include, but are not limited to, the number of visitors, the duration of visits, the number of users participating in the marketing event, the number of users who met the criteria, and the number of users who claimed the prize. These multiple risk indicators are subject to certain constraints; for example, the number of users participating in the marketing event is usually greater than the number of users who met the criteria, and the number of users participating in the marketing event is usually greater than the number of users who claimed the prize.
[0041] By identifying risk indicators and the constraints between them, risk indicators of data anomalies can be identified, laying the foundation for anomaly identification in marketing events.
[0042] Step S204: Extract the indicator data corresponding to each risk indicator from the multi-source heterogeneous data.
[0043] In step S204, the indicator data corresponding to the risk indicator is the specific value of that indicator, for example, in Figure 3 In the schematic diagram of the risk identification interface shown, the number of visitors corresponds to the indicator data of 304.
[0044] Step S205: Based on the constraint relationship between multiple risk indicators and the indicator data, determine the abnormal risk indicators from multiple risk indicators.
[0045] In step S205, the abnormal risk indicator is a key indicator for risk analysis of marketing events. This abnormal risk indicator can be an indicator whose data shows anomalies, for example, in... Figure 3 In the risk identification interface diagram shown, the number of pageviews is usually greater than the number of users who registered, but... Figure 3 If the number of page views is 0, but the number of users who registered is 8,043,367, then the number of users who registered is an anomaly, and this indicator is considered an anomaly risk indicator.
[0046] Based on the scheme defined in steps S201 to S205 above, it can be seen that in this embodiment, heterogeneous graphs and graph attention mechanisms are used to analyze multi-source heterogeneous data to determine indicators for identifying marketing event risks, making marketing event risk identification possible. In this embodiment, multi-source heterogeneous data during the marketing event process is collected in real time, including not only user behavior data but also marketing result data and transaction data generated by the marketing event. A multi-source data fusion approach is used to identify marketing event risks, improving the accuracy of risk identification. Furthermore, in this embodiment, feature extraction from multi-source heterogeneous data is achieved through heterogeneous graphs and graph attention analysis. This approach considers the strong correlation between risk indicators, thereby improving the accuracy of marketing event risk identification.
[0047] The specific implementation process of the method provided in the embodiments of this application is described below.
[0048] In some embodiments, before identifying risks in marketing events, the data processing system first needs to collect multi-source heterogeneous data. Specifically, the data processing system collects user behavior data of participating users by setting up tracking tools in the marketing platform; collects marketing result data corresponding to the marketing event from the data warehouse through the data warehouse exchange interface; and collects transaction data corresponding to the marketing event from the transaction platform through the transaction interface.
[0049] For example, the event tracking tool is set at the location of the functional script that implements the marketing event in the marketing platform. The data processing system uses a unified event tracking tool to collect user behavior data. The data warehouse is used to store the result data of the marketing event. The data processing system integrates the marketing result data of various marketing events from various financial institutions through the data warehouse exchange interface. The data processing system obtains the transaction data in the marketing event from various transaction platforms through a unified transaction interface.
[0050] In this embodiment of the application, by collecting heterogeneous data from multiple sources, the completeness and accuracy of the data are ensured, laying the foundation for the accurate identification of risks in subsequent marketing events.
[0051] In some embodiments, after acquiring multi-source heterogeneous data, the data processing system constructs a heterogeneous graph corresponding to the multi-source heterogeneous data. Specifically, the data processing system performs data fusion on the multi-source heterogeneous data to obtain fused data; then, it extracts entities corresponding to marketing events and the relationships between entities from the fused data; and finally, it constructs a heterogeneous graph corresponding to the marketing events based on the entities and the relationships between entities.
[0052] In the above embodiments, the heterogeneous graph includes nodes and edges. Nodes are used to represent entities in a marketing event, where entities may include users, activities, channels, products, etc. Edges are used to represent relationships between entities, including event participation relationships, transaction relationships, and attribution relationships.
[0053] For example, after obtaining multi-source heterogeneous data, the data processing system first performs data preprocessing, such as data cleaning, deduplication, transformation, and normalization, to ensure data quality and provide accurate data input for subsequent risk analysis of marketing events. Next, the data processing system constructs a heterogeneous graph of "user-activity-channel-product" based on the preprocessed data. Nodes in this graph include user attributes, activity information, channel identifiers, product types, etc.; edges represent participation relationships, transaction relationships, affiliation relationships, etc.
[0054] In some embodiments, the data processing system parses and filters user behavior data through a data processing engine to obtain first data; performs format conversion processing on marketing result data and transaction data to obtain second data; and merges the first data and second data using the user identifiers of participating users and a preset time window to obtain fused data.
[0055] For example, the data processing system can adopt a stream-batch integrated data processing architecture based on Apache Flink. For real-time event tracking data, it is accessed through a Kafka message queue and parsed and filtered in real time by Flink. For batch business data, it is incrementally extracted from a data warehouse (such as Hive) through scheduled tasks and converted into a unified Avro format. The two types of data are dynamically joined in the Flink job using user representation and time window as keys to achieve real-time fusion. Furthermore, in this embodiment, the data processing system also constructs a real-time metric statistics layer based on in-memory computing (such as Redis). The fused data processed by Flink is written to an OLAP database (such as ClickHouse) for historical queries, and key risk indicators (such as the number of times the same IP participates per unit time) are calculated in real time and updated to Redis. This allows the real-time monitoring and early warning unit to directly read the indicator values in Redis for comparison, decoupling millisecond-level response from complex analysis.
[0056] In some embodiments, after constructing the heterogeneous graph, the data processing system can determine risk indicators and the constraints between them based on the heterogeneous graph. Specifically, in response to an indicator selection instruction, the data processing system selects a first type of indicator and a second type of indicator from the indicator library, then obtains the graph data corresponding to the first type of indicator and the second type of indicator from the heterogeneous graph; and then inputs the graph data into a graph attention network to obtain the constraints between multiple risk indicators output by the graph attention network.
[0057] In the above embodiments, multiple risk indicators include a first type of indicator and a second type of indicator. The first type of indicator consists of key indicators necessary for risk assessment of all marketing events, while the second type of indicator matches the risk assessment needs corresponding to the marketing events. The second type of indicator is optional; for example, users can select indicators from a pool of candidate indicators through the display interface of the data processing system. A graph attention network is used to determine the feature representation of each node in a heterogeneous graph under the influence of its neighboring nodes based on graph data, and to calculate the importance of different risk indicators to risk propagation through an attention mechanism, thereby obtaining constraint relationships.
[0058] In this embodiment, the data processing system integrates internal activity data, external compliance public opinion data, and customer behavior relationship data by constructing a multi-source heterogeneous data fusion network, and automatically extracts and enhances risk features using graph neural networks and attention mechanisms. Specifically, after constructing the heterogeneous graph, the data processing system inputs the graph data of the heterogeneous graph into a multi-layer graph attention network. This graph attention network learns the feature representation of each node under the influence of its neighboring nodes, and calculates the importance weights of different edges (relationships) to risk propagation through the attention mechanism, outputting an enhanced risk feature vector for each marketing event. This vector integrates global graph structure information and local relational context.
[0059] In some embodiments, after determining the constraint relationship between multiple risk indicators, the data processing system obtains the initial risk threshold corresponding to the marketing event; and inputs the indicator data corresponding to the multiple risk indicators into the risk assessment model to obtain the initial risk score output by the risk assessment model; and then, based on the initial risk score and the initial risk threshold, the abnormal risk indicators can be determined.
[0060] In this embodiment, the data processing system employs an adaptive threshold model based on reinforcement learning (i.e., the aforementioned risk assessment model) to adjust the risk threshold in real time according to the context of the marketing event, thereby achieving accurate early warning. Simultaneously, the risk assessment model is integrated to predict risk trends.
[0061] In some embodiments, the data processing system acquires external environmental data corresponding to the marketing event, and then determines a first adjustment coefficient for adjusting the initial risk score and a second adjustment coefficient for adjusting the initial risk threshold based on the relationship between the environmental data and the adjustment coefficient; then adjusts the initial risk score and the initial risk threshold using the first adjustment coefficient and the second adjustment coefficient respectively to obtain the target risk score and the target risk threshold; finally, determines the abnormal risk indicator based on the relative value between the target risk score and the target risk threshold.
[0062] For example, the data processing system inputs enhanced feature vectors (i.e., risk indicators) into a risk assessment model (such as a deep neural network) to obtain an initial risk score. Simultaneously, a reinforcement learning agent continuously observes environmental conditions (i.e., the aforementioned external environmental data), such as the current activity type, market popularity index, and historical warning accuracy rate. Based on the current state, the agent outputs an adjustment coefficient for the initial risk score and a warning threshold offset. For instance, during peak marketing periods, the threshold for judging "suspected fraudulent activities" is automatically lowered to increase monitoring sensitivity. If the adjusted target risk score exceeds the target risk threshold, a warning signal is triggered, and the environmental state and result of this decision are recorded as experience for the agent's next learning step.
[0063] In some embodiments, after determining the abnormal risk indicators based on the relative value between the target risk score and the target risk threshold, the data processing system performs attribution analysis on the indicator data to determine the contribution of multiple risk indicators to the target risk score; then, it obtains the target risk indicators whose contribution is within a preset range, and determines the risk transmission path corresponding to the marketing event based on the target risk indicators.
[0064] In this embodiment, the data processing system can be equipped with an interpretable artificial intelligence (XAI) attribution engine to generate a risk cause analysis report simultaneously when an early warning is issued. Through counterfactual reasoning and feature importance visualization, the risk transmission path and key responsibility nodes can be clearly displayed.
[0065] Specifically, after an alert is triggered, the attribution engine of the data processing system is activated. This attribution engine uses methods such as ensemble gradient or SHAP to calculate the contribution of input features (including original features and implicit features learned by the graph network) to the high-risk score. The data processing system automatically generates a "Risk Contribution Report," displaying the top-N risk factors in the form of a bar chart or heatmap, such as "Users from XX region participated collectively through the YY channel." Then, the data processing system activates the counterfactual reasoning module to simulate modifying the value of a certain high-risk feature (for example, assuming that the number of participants through this channel decreased by 30%), recalculates the risk score, and visually demonstrates the degree of impact of the change in this factor on the risk, providing a quantitative basis for intervention.
[0066] In some embodiments, after determining abnormal risk indicators from multiple risk indicators based on the constraint relationships between multiple risk indicators and the indicator data, the data processing system also obtains marketing attribute data corresponding to the marketing event, retrieves at least one risk handling strategy corresponding to a historical marketing event from the strategy knowledge base based on the marketing attribute data, and displays the risk handling strategy.
[0067] In the above embodiments, at least one historical marketing event is a marketing event whose similarity to the marketing attribute data of the marketing event is greater than a similarity threshold. The data processing system combines a risk case library and a business rule library to provide personalized risk mitigation suggestions and uses graph computing technology to proactively detect abnormal behavior patterns and potential fraud groups, transforming passive response into proactive defense.
[0068] For example, based on the attribution results (such as the risk mainly stemming from a loophole in the rules of a specific channel), the data processing system retrieves similar historical cases and handling solutions from the strategy knowledge base. Then, it pushes a warning message to the risk administrator through the front-end interface, along with an attribution visualization report and 1-3 handling suggestions, such as, "It is recommended to immediately review the reward distribution rules of channel XX and verify the transactions of suspected abnormal user groups." Simultaneously, the data processing system also uses graph algorithms to perform community discovery on user groups in the current marketing event, identifying potential "groups" with close connections and highly similar behavioral patterns, marking them, and prompting for in-depth verification.
[0069] In addition, the risk administrator's actions and final results (whether the risk is confirmed as real and whether the action is effective) are fed back to the data processing system. This feedback can be used as new labeled data to periodically retrain the risk assessment model and reinforcement learning agent, thereby achieving continuous iterative optimization of the system's risk identification and early warning capabilities.
[0070] This concludes the introduction of the methods provided in the embodiments of this application.
[0071] As described above, this embodiment assesses and warns of marketing event management risks for financial institutions. Based on a large amount of collected user behavior data, user participation data, achievement data, and prize-winning data, it calculates and assesses marketing activities with management risks and provides early warning analysis, helping financial institutions avoid compliance risks. Furthermore, this embodiment employs integrated big data processing and statistical analysis methods in the financial field to construct a multi-indicator risk assessment model. This model comprehensively considers various effectiveness data of financial institutions' marketing events to achieve a comprehensive assessment of event activity risks. In this embodiment, a user-friendly interface allows non-technical financial institution personnel to easily use the risk assessment and early warning system. The data processing system, through simplified operation procedures and intuitive feedback mechanisms, enables personnel to quickly identify and respond to potential risks, improving the system's usability and accessibility. In this embodiment, by monitoring the data flow of financial institutions' marketing events in real time and combining it with a dynamic early warning mechanism, personnel and technicians can promptly capture risk signals and take corresponding preventative measures. This real-time approach significantly improves the bank's response speed and handling capabilities to risks.
[0072] Based on the data processing method provided in the above embodiments, this application also provides specific implementations of a data processing apparatus. Please refer to the following embodiments.
[0073] First see Figure 4 The data processing device 400 provided in this application embodiment includes: a data acquisition module 401, a graph construction module 402, an indicator determination module 403, a data extraction module 404, and a risk identification module 405.
[0074] The data acquisition module 401 is used to collect multi-source heterogeneous data generated during the marketing event. The multi-source heterogeneous data includes at least user behavior data of users participating in the marketing event, marketing result data corresponding to the marketing event, and transaction data corresponding to the marketing event. Graph construction module 402 is used to construct heterogeneous graphs corresponding to multi-source heterogeneous data; The indicator determination module 403 is used to perform graph attention analysis on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraint relationships between multiple risk indicators. Data extraction module 404 is used to extract indicator data corresponding to each risk indicator from multi-source heterogeneous data; The risk identification module 405 is used to identify abnormal risk indicators from multiple risk indicators based on the constraint relationship between multiple risk indicators and indicator data.
[0075] In some embodiments, the data acquisition module is specifically used to collect user behavior data of participating users by setting up tracking tools in the marketing platform, wherein the tracking tools are set at the location of the functional script that implements the marketing event in the marketing platform; to collect marketing result data corresponding to the marketing event from the data warehouse through the data warehouse exchange interface, wherein the data warehouse is used to store the result data of the marketing event; and to collect transaction data corresponding to the marketing event from the transaction platform through the transaction interface.
[0076] In some embodiments, the graph construction module includes: a data fusion module, a relationship extraction module, and a first construction module. The data fusion module is used to fuse multi-source heterogeneous data to obtain fused data; the relationship extraction module is used to extract entities corresponding to marketing events and the relationships between entities from the fused data; the first construction module is used to construct a heterogeneous graph corresponding to the marketing event based on the entities and the relationships between entities. The heterogeneous graph includes nodes and edges, where nodes represent entities in the marketing event, edges represent relationships between entities, and relationships include event participation relationships, transaction relationships, and attribution relationships.
[0077] In some embodiments, the data fusion module is specifically used to parse and filter user behavior data through a data processing engine to obtain first data; to perform format conversion processing on marketing result data and transaction data to obtain second data; and to fuse the first data and second data with the user identifier of the participating user and a preset time window to obtain fused data.
[0078] In some embodiments, multiple risk indicators include a first type of indicator and a second type of indicator. Specifically, the indicator determination module is used to select the first type of indicator and the second type of indicator from the indicator library in response to the indicator selection instruction. The first type of indicator is a key indicator necessary for risk assessment of all marketing events, and the second type of indicator is an indicator that matches the risk assessment needs corresponding to the marketing event. Graph data corresponding to the first type of indicator and the second type of indicator is obtained from the heterogeneous graph. The graph data is input into a graph attention network to obtain the constraint relationship between multiple risk indicators output by the graph attention network. The graph attention network is used to determine the feature representation of each node in the heterogeneous graph under the influence of its neighboring nodes based on the graph data, and to calculate the importance of different risk indicators to risk propagation through an attention mechanism to obtain the constraint relationship.
[0079] In some embodiments, the risk identification module includes: a threshold acquisition module, a score acquisition module, and a first identification module. The threshold acquisition module is used to acquire an initial risk threshold corresponding to the marketing event; the score acquisition module is used to input indicator data corresponding to multiple risk indicators into a risk assessment model to obtain an initial risk score output by the risk assessment model; and the first identification module is used to determine abnormal risk indicators based on the initial risk score and the initial risk threshold.
[0080] In some embodiments, the first identification module is specifically used to acquire external environmental data corresponding to the marketing event; determine a first adjustment coefficient for adjusting the initial risk score and a second adjustment coefficient for adjusting the initial risk threshold based on the relationship between the environmental data and the adjustment coefficient; adjust the initial risk score and the initial risk threshold using the first adjustment coefficient and the second adjustment coefficient respectively to obtain the target risk score and the target risk threshold; and determine the abnormal risk indicator based on the relative value between the target risk score and the target risk threshold.
[0081] In some embodiments, the data processing apparatus further includes: an attribution analysis module, configured to perform attribution analysis on the indicator data after determining the abnormal risk indicators based on the relative value between the target risk score and the target risk threshold, determine the contribution of multiple risk indicators to the target risk score; obtain the target risk indicators whose contribution is within a preset range; and determine the risk transmission path corresponding to the marketing event based on the target risk indicators.
[0082] In some embodiments, the data processing apparatus further includes: a strategy determination module, configured to: obtain marketing attribute data corresponding to a marketing event after determining abnormal risk indicators from multiple risk indicators based on the constraint relationship between multiple risk indicators and the indicator data; obtain risk handling strategies corresponding to at least one historical marketing event from a strategy knowledge base based on the marketing attribute data, wherein at least one historical marketing event is a marketing event whose similarity to the marketing attribute data of the marketing event is greater than a similarity threshold; and display the risk handling strategies.
[0083] The various modules of the data processing apparatus provided in the embodiments of this application can achieve Figure 2 The functions of each step in the provided data processing method, and their corresponding technical effects, will not be elaborated here for the sake of brevity.
[0084] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.
[0085] The electronic device 500 may include a processor 501 and a memory 502 storing computer program instructions.
[0086] Specifically, the processor 501 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.
[0087] Memory 502 may include mass storage for data or instructions. For example, and not limitingly, memory 502 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 502 may include removable or non-removable (or fixed) media. Where appropriate, memory 502 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 502 is non-volatile solid-state memory.
[0088] In certain embodiments, the memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Thus, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this application.
[0089] The processor 501 implements any of the data processing methods described in the above embodiments by reading and executing computer program instructions stored in the memory 502.
[0090] In some examples, the electronic device 500 may also include a communication interface 503 and a bus 504. For example, Figure 5 As shown, the processor 501, memory 502, and communication interface 503 are connected through bus 504 and complete communication with each other.
[0091] The communication interface 503 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.
[0092] Bus 504 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not as a limitation, bus 504 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 504 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.
[0093] For example, the electronic device 500 can be a mobile phone, tablet computer, laptop computer, handheld computer, in-vehicle electronic device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc.
[0094] The electronic device 500 can execute the data processing method in the embodiments of this application, thereby achieving the combination Figure 2 Describe the data processing methods.
[0095] In addition, in conjunction with the data processing methods in the above embodiments, this application also provides a computer-readable storage medium for implementation. The computer-readable storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the data processing methods in the above embodiments. Examples of computer-readable storage media include non-transitory computer-readable storage media, such as portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, etc.
[0096] This application also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, they implement any of the data processing methods described in the above embodiments.
[0097] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.
[0098] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.
[0099] This application also provides a computer program product, including a computer program, which, when executed, implements any of the data processing methods described in the above embodiments.
[0100] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.
[0101] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.
[0102] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.
Claims
1. A data processing method, characterized in that, include: Collect multi-source heterogeneous data generated during the marketing event, including at least user behavior data of participating users in the marketing event, marketing result data corresponding to the marketing event, and transaction data corresponding to the marketing event. Construct a heterogeneous graph corresponding to the multi-source heterogeneous data; Graph attention analysis is performed on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraint relationships between the multiple risk indicators. Extract indicator data corresponding to each of the risk indicators from the multi-source heterogeneous data; Based on the constraints between the multiple risk indicators, and the indicator data, abnormal risk indicators are determined from the multiple risk indicators.
2. The method according to claim 1, characterized in that, The collection of multi-source heterogeneous data generated during the marketing event includes: User behavior data of participating users is collected by setting up tracking tools in the marketing platform, wherein the tracking tools are set at the location of the functional script that implements the marketing event in the marketing platform; Marketing result data corresponding to the marketing event is collected from the data warehouse through a data warehouse exchange interface, wherein the data warehouse is used to store the result data of the marketing event; The transaction data corresponding to the marketing event is collected from the transaction platform through the transaction interface.
3. The method according to claim 2, characterized in that, The construction of the heterogeneous graph corresponding to the multi-source heterogeneous data includes: The multi-source heterogeneous data is fused to obtain fused data; Extract the entities corresponding to the marketing events and the relationships between the entities from the fused data; Based on the entities and the relationships between them, a heterogeneous graph corresponding to the marketing event is constructed. The heterogeneous graph includes nodes and edges. The nodes are used to represent entities in the marketing event, and the edges are used to represent the relationships between the entities. The relationships include event participation relationships, transaction relationships, and affiliation relationships.
4. The method according to claim 3, characterized in that, The process of fusing the multi-source heterogeneous data to obtain fused data includes: The user behavior data is parsed and filtered by the data processing engine to obtain the first data. The marketing results data and the transaction data are converted into a format to obtain the second data. The first data and the second data are fused using the user identifiers of the participating users and a preset time window to obtain the fused data.
5. The method according to claim 3, characterized in that, The multiple risk indicators include a first type of indicator and a second type of indicator. The step of performing graph attention analysis on the graph data corresponding to the heterogeneous graph to determine the multiple risk indicators and the constraint relationships between them includes: In response to the indicator selection instruction, select the first type of indicators and the second type of indicators from the indicator library, wherein the first type of indicators are key indicators necessary for risk assessment of all marketing events, and the second type of indicators are indicators that match the risk assessment needs corresponding to the marketing event. Obtain graph data corresponding to the first type of index and the second type of index from the heterogeneous graph; The graph data is input into a graph attention network to obtain the constraint relationships between multiple risk indicators output by the graph attention network. The graph attention network is used to determine the feature representation of each node in the heterogeneous graph under the influence of neighboring nodes based on the graph data, and to calculate the importance of different risk indicators to risk propagation through an attention mechanism to obtain the constraint relationships.
6. The method according to claim 1, characterized in that, The method of determining abnormal risk indicators from the multiple risk indicators based on the constraint relationships between them and the indicator data includes: Obtain the initial risk threshold corresponding to the marketing event; Input the indicator data corresponding to multiple risk indicators into the risk assessment model to obtain the initial risk score output by the risk assessment model; The abnormal risk indicator is determined based on the initial risk score and the initial risk threshold.
7. The method according to claim 6, characterized in that, The step of determining the abnormal risk indicator based on the initial risk score and the initial risk threshold includes: Obtain the external environment data corresponding to the marketing event; Based on the relationship between environmental data and adjustment coefficients, a first adjustment coefficient is determined to adjust the initial risk score, and a second adjustment coefficient is determined to adjust the initial risk threshold. The initial risk score and the initial risk threshold are adjusted using the first adjustment coefficient and the second adjustment coefficient, respectively, to obtain the target risk score and the target risk threshold; The abnormal risk indicator is determined based on the relative value between the target risk score and the target risk threshold.
8. The method according to claim 7, characterized in that, After determining the abnormal risk indicator based on the relative value between the target risk score and the target risk threshold, the method further includes: Attribution analysis is performed on the indicator data to determine the contribution of multiple risk indicators to the target risk score; Obtain the target risk indicator whose contribution level is within a preset range; The risk transmission path corresponding to the marketing event is determined based on the target risk indicators.
9. The method according to claim 8, characterized in that, After determining the abnormal risk indicators from the multiple risk indicators based on the constraint relationships between them and the indicator data, the method further includes: Obtain the marketing attribute data corresponding to the marketing event; Based on the marketing attribute data, obtain at least one risk handling strategy corresponding to a historical marketing event from the strategy knowledge base, wherein the at least one historical marketing event is a marketing event whose similarity to the marketing attribute data of the marketing event is greater than a similarity threshold; The risk management strategy described herein is presented.
10. A data processing apparatus, characterized in that, include: The data acquisition module is used to collect multi-source heterogeneous data generated during the marketing event. The multi-source heterogeneous data includes at least user behavior data of participating users in the marketing event, marketing result data corresponding to the marketing event, and transaction data corresponding to the marketing event. The graph construction module is used to construct a heterogeneous graph corresponding to the multi-source heterogeneous data; The indicator determination module is used to perform graph attention analysis on the graph data corresponding to the heterogeneous graph to determine multiple risk indicators and the constraint relationships between the multiple risk indicators. The data extraction module is used to extract indicator data corresponding to each of the risk indicators from the multi-source heterogeneous data; A risk identification module is used to determine abnormal risk indicators from the multiple risk indicators based on the constraint relationship between the multiple risk indicators and the indicator data.
11. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements the data processing method as described in any one of claims 1-9.
12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the data processing method as described in any one of claims 1-9.
13. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device is able to perform the data processing method as described in any one of claims 1-9.