Credit information pushing method and device based on data driving
By constructing credit user profiles and customer relationship graphs, combined with data indicator warehouses and push models, the problem of low information management efficiency in credit business was solved, enabling precise targeted delivery of credit information and automated decision-making, thereby improving overall work efficiency.
Patent Information
- Application Number
- CN202511729520.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-24
- Publication Date
- 2026-02-10
AI Technical Summary
In credit operations, account managers face problems of low efficiency in managing massive amounts of customer information and insufficient accuracy in decision-making. The existing task-driven work model is prone to delays in critical tasks.
User profiles and customer relationship graphs are built based on credit users' user data. Information items to be pushed are obtained through a data indicator warehouse, and accurate credit information to be pushed is generated using a push model. Automated decision-making is achieved by combining a decision tree model.
It enables precise and targeted delivery of credit information, improves work efficiency, avoids errors in human judgment, and ensures the timeliness and importance of information.
Smart Images
Figure CN121504595A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, and in particular to a data-driven credit information push method and apparatus. Background Technology
[0002] Credit business is a core business of commercial banks, characterized by rapid environmental changes, complex rules, and high compliance requirements. With the continuous growth in the number of customers, account managers are managing a large number of clients per person, resulting in heavy daily management workloads and facing the dual challenges of management efficiency and decision-making accuracy.
[0003] Currently, credit operations primarily rely on a human-centric, "task-driven" work model. Specifically, account managers arrange their daily work based on personal experience, paper-based or simple electronic to-do lists. All customer information review, process management, rule-based judgment, and report writing must be done manually. However, account managers must manually sift through massive amounts of customer information and to-do items, making them highly susceptible to delays in critical tasks due to forgetfulness or misjudgment of priorities, resulting in low work efficiency.
[0004] Therefore, how to improve work efficiency has become an urgent problem to be solved in this field. Summary of the Invention
[0005] This application provides a data-driven credit information push method and apparatus, with the aim of improving work efficiency.
[0006] To achieve the above objectives, this application provides the following technical solution:
[0007] The first aspect of this application provides a data-driven credit information push method, including:
[0008] Collect user data of credit users and the credit customers associated with those credit users;
[0009] A credit user profile is determined based on the credit user's user data.
[0010] By analyzing the credit customers associated with the credit users, a credit customer relationship graph is obtained;
[0011] The data indicator warehouse retrieves information items to be pushed; these information items are important information that the credit user needs to pay attention to or process.
[0012] The credit user profile, the customer relationship graph, and the information items to be pushed are input into the push model to obtain the credit information to be pushed.
[0013] The credit information to be pushed will be sent to the credit user.
[0014] Optionally, determining the credit user profile based on the credit user's user data includes:
[0015] The similarity between the user data of the credit user and the user data of other credit users is calculated using a dynamic time warping algorithm; the user data of other credit users refers to the user data of other credit users besides the user data of the credit user.
[0016] The credit users are classified according to the similarity using a clustering algorithm to obtain multiple groups of credit users;
[0017] By analyzing the user data of multiple groups of credit users and the credit users themselves, a credit user profile is obtained.
[0018] Optionally, the step of analyzing the credit customers associated with the credit user to obtain a credit customer relationship graph includes:
[0019] The credit customers were analyzed to obtain their relationships.
[0020] A heterogeneous graph network is constructed based on the relationships between the credit customers; the heterogeneous graph network includes nodes and edges; the nodes are the credit customers; the edges are the relationships between the credit customers;
[0021] The heterogeneous graph network is used to extract features to obtain business features;
[0022] The business characteristics are input into a graph attention network model to obtain a credit customer relationship graph.
[0023] Optionally, before obtaining the information items to be pushed from the data metric warehouse, the method further includes:
[0024] Obtain historical credit process data;
[0025] Feature data is obtained by extracting features from the historical credit process data;
[0026] The feature data is preprocessed to obtain preprocessed feature data;
[0027] The preprocessed feature data is input into the vector machine model to obtain key indicators;
[0028] The key indicators are classified to obtain information items to be pushed, and the information items to be pushed are stored in the data indicator warehouse.
[0029] Optional, also includes:
[0030] The credit information to be pushed is input into a decision tree model to obtain a processing strategy;
[0031] The processing strategy and the credit information to be pushed are sent to the credit user.
[0032] A second aspect of this application provides a data-driven credit information push device, comprising:
[0033] The data collection unit is used to collect user data of credit users and credit customers associated with the credit users;
[0034] A determining unit is used to determine a credit user profile based on the credit user's user data;
[0035] The analysis unit is used to analyze the credit customers associated with the credit user to obtain a credit customer relationship graph;
[0036] The acquisition unit is used to acquire information items to be pushed from the data indicator warehouse; the information items to be pushed are important information that the credit user needs to pay attention to or process.
[0037] The input unit is used to input the credit user profile, the customer relationship graph, and the information items to be pushed into the push model to obtain the credit information to be pushed.
[0038] The push unit is used to push the credit information to be pushed to the credit user.
[0039] Optionally, the determining unit is specifically used for:
[0040] The similarity between the user data of the credit user and the user data of other credit users is calculated using a dynamic time warping algorithm; the user data of other credit users refers to the user data of other credit users besides the user data of the credit user.
[0041] The credit users are classified according to the similarity using a clustering algorithm to obtain multiple groups of credit users;
[0042] By analyzing the user data of multiple groups of credit users and the credit users themselves, a credit user profile is obtained.
[0043] Optionally, the analysis unit is specifically used for:
[0044] The credit customers were analyzed to obtain their relationships.
[0045] A heterogeneous graph network is constructed based on the relationships between the credit customers; the heterogeneous graph network includes nodes and edges; the nodes are the credit customers; the edges are the relationships between the credit customers;
[0046] The heterogeneous graph network is used to extract features to obtain business features;
[0047] The business characteristics are input into a graph attention network model to obtain a credit customer relationship graph.
[0048] Optional, also includes:
[0049] The data acquisition unit is used to acquire historical credit process data;
[0050] The extraction unit is used to extract features from the historical credit process data to obtain feature data.
[0051] The preprocessing unit is used to preprocess the feature data to obtain preprocessed feature data;
[0052] The data input unit is used to input the preprocessed feature data into the vector machine model to obtain key indicators;
[0053] The classification unit is used to classify the key indicators, obtain the information items to be pushed, and store the information items to be pushed in the data indicator warehouse.
[0054] Optional, also includes:
[0055] An information input unit is used to input the credit information to be pushed into a decision tree model to obtain a processing strategy;
[0056] An information push unit is used to push the processing strategy and the credit information to be pushed to the credit user.
[0057] The technical solution provided in this application determines credit user profiles based on user data of credit users; analyzes the credit customers associated with credit users to obtain a credit customer relationship graph; obtains information items to be pushed from a data indicator warehouse; inputs the credit user profile, customer relationship graph, and information items to be pushed into a push model to obtain credit information to be pushed; and pushes the credit information to be pushed to credit users. After the credit user profile, customer relationship graph, and information items to be pushed are input into the push model, the model generates credit information to be pushed with priority. It can identify and filter the key information that is most relevant to the user and that requires the most attention or processing, ensuring that the pushed content is both accurate and prioritized. In this way, not only is the orderly arrangement and targeted push of tasks achieved, but errors that may be caused by manual judgment are also effectively avoided, thereby significantly improving overall work efficiency. Attached Figure Description
[0058] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0059] Figure 1 A flowchart illustrating a data-driven credit information push method provided in this application embodiment;
[0060] Figure 2 This is a schematic diagram of the architecture of a data-driven credit information push device provided in an embodiment of this application. Detailed Implementation
[0061] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0062] In this application, 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 limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0063] like Figure 1 The flowchart shown is a data-driven credit information push method provided in this application embodiment, including the following steps:
[0064] S101: Collect user data of credit users and credit customers associated with credit users.
[0065] Optionally, credit users can be employees of credit institutions, or individuals or businesses that have established a formal relationship with credit institutions (such as banks, loan companies, etc.).
[0066] Among them, the user data of credit users includes, but is not limited to, user identity, institutional information, personnel permissions, and behavioral patterns.
[0067] It should be noted that the information (including but not limited to user identity, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.
[0068] S102: Determine credit user profiles based on credit user data.
[0069] Understandably, determining credit user profiles based on credit user data involves re-organizing all credit users and integrating multiple user roles (a user's unique system identity representing the user's category group), thereby creating credit user profiles based on these user roles.
[0070] Optionally, in another embodiment of this application, the specific implementation of step S102 includes processes A1 to A3.
[0071] A1: Calculate the similarity between user data of credit users and user data of other credit users using the dynamic time warping algorithm.
[0072] Among them, the user data of other credit users refers to the user data of other credit users besides the user data of other credit users.
[0073] Specifically, using the work line as the horizontal axis, the functional identity as the vertical axis, and the organizational level as the cross-section, and combining the dynamic time warping algorithm, the similarity of the operation time series between different credit users is calculated, that is, the similarity between the user data of credit users and the user data of other credit users is calculated.
[0074] A2: Use clustering algorithms to classify credit users according to similarity to obtain multiple groups of credit users.
[0075] Optionally, clustering algorithms include, but are not limited to, the K-Means algorithm.
[0076] Understandably, by using similarity as a feature, the K-Means algorithm can be used to group credit users into different groups, thus obtaining multiple groups of credit users.
[0077] A3: Analyze the user data of multiple groups of credit users to obtain credit user profiles.
[0078] This involved analyzing user data from multiple credit user groups and conducting multi-dimensional time-series analysis on credit user groups to extract key data features and construct fine-grained user profiles. Through a comprehensive overhaul of the entire credit user system, we integrated and designed over 300 new user roles, thereby creating a new generation of three-dimensional credit user profiles.
[0079] In addition, principal component analysis (PCA) can be used to reduce the dimensionality of high-dimensional user data, ensuring that key features of the user data are preserved while reducing computational complexity, thus laying the foundation for further data mining.
[0080] S103: Analyze the credit customers associated with the credit users to obtain a credit customer relationship graph.
[0081] This involves analyzing the credit customers associated with credit users, examining the various relationships between credit customers, and thus obtaining a credit customer relationship map.
[0082] Optionally, in another embodiment of this application, the specific implementation of step S103 includes processes B1 to B4.
[0083] B1: Analyze credit customers to obtain their relationships.
[0084] Optionally, heterogeneous graph neural networks can be used to analyze credit customers and obtain their relationships.
[0085] B2: Construct a heterogeneous graph network based on the relationships between credit customers.
[0086] The heterogeneous graph network consists of nodes and edges; nodes represent credit customers; and edges represent the relationships between credit customers.
[0087] Specifically, in heterogeneous graph networks, in addition to basic relationships such as primary and secondary credit offices and supervisors, there are also implicit relationships such as group members and those responsible for approval business.
[0088] B3: Use a multi-hop path approach to extract features from heterogeneous graph networks to obtain business features.
[0089] Specifically, information is propagated and aggregated from nodes in a heterogeneous graph network through multi-hop paths to extract business-related features and obtain business features.
[0090] B4: Input the business features into the graph attention network model to obtain the credit customer relationship graph.
[0091] Understandably, by inputting business characteristics into a graph attention network model and weighting the attention of various relationship nodes, the model can filter out the relationship types that contribute most significantly to business risk prediction, thereby extracting a variety of commonly used customer relationship types (e.g., 17 commonly used customer relationship types) to comprehensively depict the customer groups associated with credit users. Based on social network analysis technology, further integration and refinement of various close relationships among customers, such as kinship, investment, guarantee, directors, and family businesses, are conducted to identify the complex relationship network structure among customers, and a credit customer relationship map is constructed based on the relationship network structure.
[0092] It can be seen that this method can construct a multi-level, multi-dimensional customer relationship network, effectively expand the coverage of credit customer group identification, and ultimately form a complete credit customer group relationship map.
[0093] S104: Retrieve information items to be pushed from the data metrics warehouse.
[0094] Among them, the information to be pushed is important information that credit users need to pay attention to or process. For example, important information includes tasks that users need to complete within a specified time, risk warnings, reminders of important matters, and changes in key operating indicators.
[0095] Optionally, before step S104, the information items to be pushed need to be pre-stored in a data indicator warehouse. These information items can be updated periodically as needed and kept up-to-date in the data indicator warehouse. In this way, relevant data can be quickly and efficiently extracted from the data indicator warehouse during subsequent push processes, ensuring accurate information transmission and timely push. Therefore, another embodiment of this application provides an information storage method, including processes C1 to C5.
[0096] C1: Obtain historical credit process data.
[0097] Historical credit process data refers to the full-process, multi-dimensional data generated and accumulated in past credit business, from customer application to loan settlement.
[0098] C2: Extract features from historical credit process data to obtain feature data.
[0099] Feature data refers to characteristics related to key indicators, including but not limited to customer credit scores, income levels, loan approval times, interest rates, repayment records, and overdue days.
[0100] C3: Perform data preprocessing on the feature data to obtain preprocessed feature data.
[0101] This includes data preprocessing of the feature data, including cleaning and standardizing the feature data to obtain preprocessed feature data.
[0102] C4: Input the preprocessed feature data into the vector machine model to obtain key indicators.
[0103] Among them, key indicators refer to the data that best reveals the likelihood of customer default.
[0104] Optionally, the vector machine model can be an SVM model.
[0105] It should be noted that the vector machine model is pre-trained based on sample feature data as input and key indicators of manually labeled sample feature data as training targets.
[0106] C5: Classify key indicators to obtain information items to be pushed, and store the information items to be pushed to the data indicator warehouse.
[0107] Understandably, key indicators are categorized, and information items to be pushed are extracted and summarized from three dimensions: due date reminders, risk warnings, and indicators of concern. The information items to be pushed are used to monitor data such as time-limited tasks for users, risk warnings, important event reminders, and changes in key operating indicators.
[0108] As can be seen, through in-depth analysis, modeling, and relationship mining of massive amounts of data (i.e., historical credit process data), we have built a data indicator warehouse covering the entire credit process, from pre-loan to post-loan. This warehouse can systematically assist users in identifying key tasks and potential risks at each stage, thereby improving the efficiency of the entire process management.
[0109] Optionally, it can integrate various pending tasks from before, during, and after the loan disbursement process, enabling targeted delivery of these tasks to loan users.
[0110] S105: Input the credit user profile, customer relationship graph and information items to be pushed into the push model to obtain the credit information to be pushed.
[0111] Among them, credit information to be pushed refers to action instructions or risk insight information with clear business objectives that are prepared to be sent to different credit users.
[0112] Optionally, the push model can be a long short-term memory network with memory. The push model is pre-trained based on sample credit user profiles, sample customer relationship graphs, and sample information items to be pushed as inputs, and sample credit information to be pushed that is manually labeled by the sample credit user profiles, sample customer relationship graphs, and sample information items to be pushed as training targets.
[0113] Understandably, credit user profiles, customer relationship graphs, and the information items to be pushed are all input into the push model. The push model will configure differentiated information indicators and customer relationships for different credit user roles, thereby driving the push engine to accurately recommend differentiated work content to personnel at different levels, roles, and lines, and ultimately generate credit information to be pushed.
[0114] In addition, we will continuously monitor the behavioral data of credit users and adjust the content pushed to ensure the relevance and effectiveness of the information push.
[0115] It should be noted that by using a long short-term memory network with memory to model user historical behavior data with time series, the push strategy can be adaptively adjusted to meet the dynamic needs of users.
[0116] S106: Push the credit information to be pushed to credit users.
[0117] Optionally, after step S106, the credit information of different users can be analyzed to generate targeted processing strategies, so that the credit recommendations or processing solutions received by each user are more in line with their credit status. Therefore, in another embodiment of this application, a processing strategy push method is provided, including processes D1 to D2.
[0118] D1: Input the credit information to be pushed into the decision tree model to obtain the processing strategy.
[0119] The processing strategy refers to the specific and actionable plan to be implemented. For example, account freezing or credit limit reduction.
[0120] Understandably, when credit information to be pushed is input into a decision tree model, the decision tree model makes a series of judgments based on the key features of the task, and finally classifies it into the most matching leaf node. Each leaf node corresponds to a specific processing strategy (such as task allocation, priority setting, or automated processing).
[0121] It should be noted that the decision tree model is trained in advance by analyzing historical data in the loan business process (i.e., historical credit information to be pushed), extracting key features such as task type, priority and user behavior, and mapping tasks to the optimal decision path.
[0122] D2: Push the processing strategy and credit information to be pushed to credit users.
[0123] Understandably, the processing strategy and pending credit information are pushed to credit users to trigger their next action. First, for clear tasks such as account anomalies, behavioral anomalies, and interest collection, automatic handling is achieved, including account freezing, credit limit deduction, and intelligent outbound calls. Second, for tasks with complex rules that require manual intervention, generative AI based on neural networks is used to automatically generate key summary information and risk warning labels, providing suggestions for the next steps for user decision-making.
[0124] In summary, after inputting credit user profiles, customer relationship graphs, and information items to be pushed into the push model, the model analyzes the data and generates prioritized credit information to be pushed. It can identify and filter the most relevant, urgent, or critical information for users, ensuring that the pushed content is both accurate and prioritized. This approach not only achieves orderly task arrangement and targeted push but also effectively avoids errors that may arise from manual judgment, thus significantly improving overall work efficiency.
[0125] like Figure 2 The diagram shown is an architectural schematic of a data-driven credit information push device provided in this application embodiment. The credit information push device includes: a collection unit 100, a determination unit 200, an analysis unit 300, an acquisition unit 400, an input unit 500, and a push unit 600.
[0126] The data collection unit 100 is used to collect user data of credit users and credit customers associated with the credit users.
[0127] The determining unit 200 is used to determine a credit user profile based on the credit user's user data.
[0128] The determining unit 200 is specifically used for: using a dynamic time warping algorithm to calculate the similarity between the user data of the credit user and the user data of other credit users; the user data of other credit users refers to the user data of other credit users besides the user data of the credit user; using a clustering algorithm to classify the credit user according to the similarity to obtain multiple groups of credit users; and analyzing the multiple groups of credit users and the user data of the credit user to obtain a credit user profile.
[0129] The analysis unit 300 is used to analyze the credit customers associated with the credit user to obtain a credit customer relationship graph.
[0130] The analysis unit 300 is specifically used for: analyzing the credit customers to obtain the relationships between them; constructing a heterogeneous graph network based on the relationships between the credit customers; the heterogeneous graph network includes nodes and edges; the nodes are the credit customers; the edges are the relationships between the credit customers; extracting features from the heterogeneous graph network using a multi-hop path method to obtain business features; and inputting the business features into a graph attention network model to obtain a credit customer relationship graph.
[0131] The acquisition unit 400 is used to acquire information items to be pushed from the data indicator warehouse; the information items to be pushed are important information that the credit user needs to pay attention to or process.
[0132] The input unit 500 is used to input the credit user profile, the customer relationship graph, and the information items to be pushed into the push model to obtain the credit information to be pushed.
[0133] The push unit 600 is used to push the credit information to be pushed to the credit user.
[0134] In summary, after inputting credit user profiles, customer relationship graphs, and information items to be pushed into the push model, the model analyzes the data and generates prioritized credit information to be pushed. It can identify and filter the most relevant, urgent, or critical information for users, ensuring that the pushed content is both accurate and prioritized. This approach not only achieves orderly task arrangement and targeted push but also effectively avoids errors that may arise from manual judgment, thus significantly improving overall work efficiency.
[0135] Combination Figure 2 The credit information push device, as shown, also includes:
[0136] The data acquisition unit is used to acquire historical credit process data.
[0137] The extraction unit is used to extract features from the historical credit process data to obtain feature data.
[0138] The preprocessing unit is used to preprocess the feature data to obtain preprocessed feature data.
[0139] The data input unit is used to input the preprocessed feature data into the vector machine model to obtain key indicators.
[0140] The classification unit is used to classify the key indicators, obtain the information items to be pushed, and store the information items to be pushed in the data indicator warehouse.
[0141] Combination Figure 2 The credit information push device, as shown, also includes:
[0142] The information input unit is used to input the credit information to be pushed into the decision tree model to obtain the processing strategy.
[0143] An information push unit is used to push the processing strategy and the credit information to be pushed to the credit user.
[0144] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. In particular, for system or system embodiments, since they are fundamentally similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. Components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.
[0145] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0146] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A data-driven credit information push method, characterized in that, include: Collect user data of credit users and the credit customers associated with those credit users; A credit user profile is determined based on the credit user's user data. By analyzing the credit customers associated with the credit users, a credit customer relationship graph is obtained; The data indicator warehouse retrieves information items to be pushed; these information items are important information that the credit user needs to pay attention to or process. The credit user profile, the customer relationship graph, and the information items to be pushed are input into the push model to obtain the credit information to be pushed. The credit information to be pushed will be sent to the credit user.
2. The method according to claim 1, characterized in that, The process of determining a credit user profile based on the credit user's user data includes: The similarity between the user data of the credit user and the user data of other credit users is calculated using a dynamic time warping algorithm; the user data of other credit users refers to the user data of other credit users besides the user data of the credit user. The credit users are classified according to the similarity using a clustering algorithm to obtain multiple groups of credit users; By analyzing the user data of multiple groups of credit users and the credit users themselves, a credit user profile is obtained.
3. The method according to claim 1, characterized in that, The analysis of the credit customers associated with the credit user to obtain a credit customer relationship graph includes: The credit customers were analyzed to obtain their relationships. A heterogeneous graph network is constructed based on the relationships between the credit customers; the heterogeneous graph network includes nodes and edges; the nodes are the credit customers; the edges are the relationships between the credit customers; The heterogeneous graph network is used to extract features to obtain business features; The business characteristics are input into a graph attention network model to obtain a credit customer relationship graph.
4. The method according to claim 1, characterized in that, Before retrieving the information items to be pushed from the data indicator warehouse, the process also includes: Obtain historical credit process data; Feature data is obtained by extracting features from the historical credit process data; The feature data is preprocessed to obtain preprocessed feature data; The preprocessed feature data is input into the vector machine model to obtain key indicators; The key indicators are classified to obtain information items to be pushed, and the information items to be pushed are stored in the data indicator warehouse.
5. The method according to claim 1, characterized in that, Also includes: The credit information to be pushed is input into a decision tree model to obtain a processing strategy; The processing strategy and the credit information to be pushed are sent to the credit user.
6. A data-driven credit information push device, characterized in that, include: The data collection unit is used to collect user data of credit users and credit customers associated with the credit users; A determining unit is used to determine a credit user profile based on the credit user's user data; The analysis unit is used to analyze the credit customers associated with the credit user to obtain a credit customer relationship graph; The acquisition unit is used to acquire information items to be pushed from the data indicator warehouse; the information items to be pushed are important information that the credit user needs to pay attention to or process. The input unit is used to input the credit user profile, the customer relationship graph, and the information items to be pushed into the push model to obtain the credit information to be pushed. The push unit is used to push the credit information to be pushed to the credit user.
7. The apparatus according to claim 6, characterized in that, The determining unit is specifically used for: The similarity between the user data of the credit user and the user data of other credit users is calculated using a dynamic time warping algorithm; the user data of other credit users refers to the user data of other credit users besides the user data of the credit user. The credit users are classified according to the similarity using a clustering algorithm to obtain multiple groups of credit users; By analyzing the user data of multiple groups of credit users and the credit users themselves, a credit user profile is obtained.
8. The apparatus according to claim 6, characterized in that, The analysis unit is specifically used for: The credit customers were analyzed to obtain their relationships. A heterogeneous graph network is constructed based on the relationships between the credit customers; the heterogeneous graph network includes nodes and edges; the nodes are the credit customers; the edges are the relationships between the credit customers; The heterogeneous graph network is used to extract features to obtain business features; The business characteristics are input into a graph attention network model to obtain a credit customer relationship graph.
9. The apparatus according to claim 6, characterized in that, Also includes: The data acquisition unit is used to acquire historical credit process data; The extraction unit is used to extract features from the historical credit process data to obtain feature data. The preprocessing unit is used to preprocess the feature data to obtain preprocessed feature data; The data input unit is used to input the preprocessed feature data into the vector machine model to obtain key indicators; The classification unit is used to classify the key indicators, obtain the information items to be pushed, and store the information items to be pushed in the data indicator warehouse.
10. The apparatus according to claim 6, characterized in that, Also includes: An information input unit is used to input the credit information to be pushed into a decision tree model to obtain a processing strategy; An information push unit is used to push the processing strategy and the credit information to be pushed to the credit user.