Credit information data management and analysis method and system

Abstract

The application is suitable for information management method technology, and provides a credit information data management and analysis method and system, credit data information is acquired; a multi-dimensional credit map is formed according to edge weight constructed according to an agent variable, a credit behavior similar community is identified according to existing data, a positioning similar crowd is obtained, and a contribution score is determined according to a white household user contribution degree; a multi-dimensional credit behavior trajectory graph is constructed according to real-time behavior data flow and platform data, a mutation node in the trajectory graph is identified, whether the behavior change constitutes a credit risk deviation signal is judged according to the mutation node; the positive or negative influence amplitude of each behavior characteristic on the final score is quantified according to the final score, updated evidence material information is acquired according to the score elements, a new score suggestion value is obtained according to the evidence material information, and the key problems that are difficult to model white household users in a traditional credit system, that the scoring result is static and rigid, and that a complaint cannot be fed back are effectively solved.

Description

Technical Field

[0001] This invention belongs to the field of information management methods and technology, and in particular relates to a method and system for managing and analyzing credit information data. Background Technology

[0002] Currently, credit scoring systems are widely used in various scenarios such as financial lending, e-commerce credit, and leasing services. Their core function is to quantitatively assess users' credit behavior to aid risk control and decision support. However, existing credit scoring systems primarily rely on users' historical credit data, such as loan records, repayment behavior, and credit card usage. This fails to effectively cover users with no credit history or minimal credit activity, resulting in a large number of potentially low-risk users being excluded from traditional financial services. Furthermore, traditional scoring models generally employ static rules or fixed weight calculation methods, making it difficult to dynamically reflect real-time changes in user behavior. They lack the ability to promptly identify and respond to credit risks caused by unforeseen events, easily leading to delayed or distorted scoring results, thus reducing the sensitivity and accuracy of risk control.

[0003] On the other hand, most current credit scoring systems are closed models, lacking transparency in the scoring process. Users typically only receive the final score and cannot understand the composition of the score or the specific impact of each behavioral factor. This not only reduces users' trust in the scoring results but also fails to meet their needs for feedback and appeals regarding scoring disputes. In the event of misjudgment or abnormal deductions, users often cannot provide valid materials to participate in score correction, affecting their normal rights in financial activities.

[0004] Therefore, there is an urgent need for an intelligent scoring method that can integrate multi-source data, is applicable to users with no credit history, supports dynamic adjustments based on behavioral changes, and has interpretability and appeal mechanisms, in order to improve the applicability, real-time responsiveness, and user participation of the scoring model, and promote the development of the credit reporting system towards a more open, fair, and trustworthy direction. Summary of the Invention

[0005] The purpose of this invention is to provide a method for managing and analyzing credit information data, which aims to solve the problems raised in the third part of the background technology.

[0006] The present invention is implemented as follows: a method for managing and analyzing credit information data, the method comprising:

[0007] Credit information is obtained through multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators.

[0008] A multidimensional credit graph is formed by constructing edge weights based on proxy variables. Similar credit behavior communities are identified based on existing data to obtain similar groups of people. Contribution scores are determined based on the contribution of new users.

[0009] A multidimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. Abrupt nodes in the trajectory map are identified, and it is determined whether the behavioral change constitutes a signal of credit risk shift based on the abrupt nodes.

[0010] The final score quantifies the positive or negative impact of each behavioral characteristic on the final score, updates the evidence materials based on the scoring elements, and then derives new suggested scores based on the evidence materials.

[0011] Preferably, the steps of constructing edge weights to form a multidimensional credit graph based on proxy variables, identifying similar credit behavior communities based on existing data, obtaining similar user groups, and determining contribution scores based on the contribution of new users specifically include:

[0012] The system identifies "white-listed" users based on credit data. These white-listed users are those with no credit history. The system obtains proxy variables based on white-listed users, including utility payment records such as water, electricity and gas, and fulfillment and return rates on e-commerce platforms.

[0013] Financial behavior capability is determined based on proxy variables, and a multi-dimensional credit graph is formed by constructing edge weights based on proxy variables. Existing data is obtained, and communities with similar credit behavior are identified based on the existing data to obtain the location of similar people.

[0014] Obtain the contribution score of existing new users, which includes continuous payment, positive reviews on e-commerce platforms, and unstable location. Determine the contribution score based on the contribution score of new users.

[0015] Preferably, the step of constructing a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, identifying abrupt change nodes in the trajectory map, and determining whether the behavioral change constitutes a signal of credit risk shift based on the abrupt change nodes specifically includes:

[0016] Acquire real-time behavioral data streams, including financial, e-commerce, and public service data, and obtain platform data by connecting to the platform through authorized API interfaces;

[0017] A multidimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. The dimensions of the trajectory map represent key credit factors, and abrupt change nodes are identified in the trajectory map.

[0018] The change in behavior is determined based on the mutation node to determine whether it constitutes a signal of credit risk deviation. If a credit risk deviation is determined, the score is adjusted down by calculating the correction factor, and the node record is retained for trend monitoring.

[0019] Preferably, the steps of quantifying the positive or negative impact of each behavioral feature on the final score based on the final score, obtaining updated evidence information based on the scoring elements, and obtaining new suggested scores based on the evidence information specifically include:

[0020] Obtain the corrected final score, quantify the positive or negative impact of each behavioral feature on the final score, and visualize the quantification results.

[0021] Obtain the scoring components, which include credit enhancement factors, credit risk factors, and neutral characteristics. Based on the scoring components, obtain updated evidentiary information, which is used for scoring appeals.

[0022] If a new suggested score is obtained based on the evidence, and the suggested score differs significantly from the original score and the evidence is credible, the score is corrected and the score correction path is recorded.

[0023] Preferably, the platform includes bank credit reporting, payment platforms, e-commerce platforms, and operators.

[0024] Another objective of this invention is to provide a credit information data management and analysis system, the system comprising:

[0025] The credit information module acquires credit data information, which is obtained from multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators.

[0026] The contribution scoring module constructs edge weights based on proxy variables to form a multi-dimensional credit graph, identifies similar credit behavior communities based on existing data, obtains similar user groups, and determines contribution scores based on the contribution of new users.

[0027] The risk identification module constructs a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, identifies abrupt change nodes in the trajectory map, and determines whether the behavioral change constitutes a signal of credit risk shift based on the abrupt change nodes.

[0028] The scoring interpretation module quantifies the positive or negative impact of each behavioral feature on the final score, obtains updated evidence information based on the scoring elements, and derives new suggested scores based on the evidence information.

[0029] Preferably, the contribution scoring module includes:

[0030] The proxy variable unit determines blank users based on credit data. Blank users are users with no credit history. Proxy variables are obtained based on blank users. Proxy variables include public utility payment records such as water, electricity, and gas, and fulfillment rates and return rates on e-commerce platforms.

[0031] The credit graph unit determines financial behavior capability based on proxy variables, constructs edge weights based on proxy variables to form a multi-dimensional credit graph, acquires existing data, identifies similar credit behavior communities based on existing data, and obtains the location of similar people.

[0032] The contribution score unit obtains the contribution score of existing new users, which includes continuous payment, positive reviews on e-commerce platforms, and unstable location. The contribution score is determined based on the contribution score of new users.

[0033] Preferably, the risk identification module includes:

[0034] The real-time behavior unit acquires real-time behavior data streams, including financial, e-commerce, and public service data, and obtains platform data by connecting to the platform through authorized API interfaces.

[0035] The behavioral trajectory graph unit constructs a multi-dimensional credit behavior trajectory graph based on real-time behavioral data streams and platform data. The trajectory graph dimensions represent key credit factors and identify abrupt change nodes in the trajectory graph.

[0036] The risk offset unit determines whether a change in behavior constitutes a credit risk offset signal based on the mutation node. If a credit risk offset is determined to exist, the score is adjusted downward by calculating a correction factor, and the node record is retained for trend monitoring.

[0037] Preferably, the scoring interpretation module includes:

[0038] The scoring interpretation unit obtains the corrected final score, quantifies the positive or negative impact of each behavioral feature on the final score, and visualizes the quantification results.

[0039] The material information update unit obtains the scoring components, which include credit enhancement factors, credit risk factors, and neutral characteristics. Based on the scoring components, it obtains updated evidence material information, which is used for scoring appeals.

[0040] The scoring correction unit obtains a new suggested scoring value based on the evidence material information. If the suggested value differs significantly from the original scoring and the evidence material information is credible, the scoring is corrected and the scoring correction path is recorded.

[0041] Preferably, the platform includes bank credit reporting, payment platforms, e-commerce platforms, and operators.

[0042] This invention provides a credit information data management and analysis method, offering an intelligent credit scoring approach that integrates multi-source credit data fusion, proxy variable modeling, graph-structured community identification, dynamic behavioral trajectory analysis, and score correction. This effectively solves key problems in traditional credit scoring systems, such as the difficulty in modeling users with no credit history, the static and rigid scoring results, and the inability to provide feedback on appeals. By introducing proxy variables such as public utility payments, e-commerce fulfillment, communication behavior, and location information, a multi-dimensional feature vector of the user is constructed. A graph-structured approach is then used to identify similar groups within the credit behavior graph, enabling behavioral classification and initial score estimation for users with no credit history. This mechanism significantly expands the applicability of the scoring system to users without credit data, improving the coverage and initial judgment capabilities of the scoring model.

[0043] During the scoring process, this invention acquires real-time behavioral data streams from platforms such as finance, e-commerce, and public services to construct a multi-dimensional credit behavior trajectory map. It employs a sliding window and anomaly detection algorithms to identify abrupt change nodes, determine whether a user exhibits credit risk deviation behavior, and dynamically adjusts the score based on the magnitude and importance of the abrupt change. The final score not only reflects stable user behavior but also embodies trends in behavioral changes.

[0044] After the scoring results are generated, the system further calls an interpretable model to quantify and visualize the contribution of each factor to the score, and supports users to submit evidence materials for appeal based on the scoring factors. After the materials are verified by OCR and semantics, a partial re-scoring process is triggered. If the suggested score changes significantly and the materials are credible, the score is automatically corrected and the score change path is recorded. Attached Figure Description

[0045] Figure 1 A flowchart illustrating a credit information data management and analysis method provided in this embodiment of the invention;

[0046] Figure 2 This is a flowchart of the steps for constructing edge weights to form a multidimensional credit graph based on proxy variables and determining contribution scores based on the contribution of new users, as provided in an embodiment of the present invention.

[0047] Figure 3 The flowchart of the steps for constructing a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, and determining whether a change in behavior constitutes a signal of credit risk shift based on abrupt change nodes, is provided in this embodiment of the invention.

[0048] Figure 4 The flowchart illustrates the steps of quantifying the positive or negative impact of each behavioral feature on the final score based on the final score, and obtaining a new suggested score value based on the evidence material information provided in this embodiment of the invention.

[0049] Figure 5An architecture diagram of a credit information data management and analysis system provided in this embodiment of the invention;

[0050] Figure 6 This is an architecture diagram of the contribution scoring module provided in an embodiment of the present invention;

[0051] Figure 7 This is an architecture diagram of the risk identification module provided in an embodiment of the present invention;

[0052] Figure 8 This is an architecture diagram of the scoring interpretation module provided in an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0054] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise stated, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.

[0055] like Figure 1 As shown, this invention provides a method for managing and analyzing credit information data, the method comprising:

[0056] S100, Obtain credit information, which is obtained from multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators.

[0057] In this step, credit data is acquired using a multi-source heterogeneous data fusion mechanism, covering multiple sources such as banks, consumer finance institutions, and telecommunications operators to improve the completeness and representativeness of user credit profiles. At the bank level, data can be accessed from traditional commercial banks and internet banks, including account transaction data, credit card usage records, repayment history, and loan contract information, with particular attention paid to quantifiable indicators such as repayment frequency, bill amount ratio, and delinquency behavior. For example, if a user has multiple small loans in the past 12 months, without delinquency, but frequently switches loan platforms, this behavior will be used as a risk boundary factor in the model.

[0058] In the consumer finance dimension, by accessing data interfaces of consumer installment platforms, micro-loan systems and e-commerce financial products, we can extract users' performance in installment shopping, cash installment, BNPL (buy now, pay later), approval rate, credit limit utilization rate and dispute records to establish a model of users' willingness to assume responsibility and financial behavior habits.

[0059] Telecommunications operator data serves as an auxiliary dimension for credit scoring, providing users with detailed call records, number usage duration, payment cycle patterns, roaming records, and location trajectories to assess their life stability and credit consistency. For example, long-term arrears in phone bills, frequent changes in mobile phone numbers, or sudden changes in call patterns will be treated as potential credit default warning factors.

[0060] S200 constructs edge weights based on proxy variables to form a multidimensional credit graph, identifies similar credit behavior communities based on existing data, obtains similar user groups, and determines contribution scores based on the contribution of new users.

[0061] In this step, edge weights are constructed based on proxy variables to form a multidimensional credit graph. To model the credit behavior of users without credit history, edge weights are first constructed based on proxy variables to form a multidimensional credit graph reflecting the relationships between user behavioral characteristics. Specifically, each user constructs a feature vector through their proxy variables, such as the continuity of utility bill payments, e-commerce fulfillment rate, device location stability, and phone bill payment habits. The graph structure constructed in this way not only preserves the behavioral similarity relationships between users but also maps the implicit behavioral trend proximity in the graph.

[0062] Next, blank users are integrated into the graph. Based on the edge weights between their proxy feature vectors and labeled users, the most similar Top-K users or communities are identified. Then, based on the historical average and fluctuations of users within the community, the fitting credit behavior trends of blank users across various dimensions are preliminarily estimated. The degree of fitting and perturbation of blank users to the feature centers of adjacent communities in the graph structure is evaluated. If their behavior is highly consistent with the core features of the community and does not cause significant structural perturbation, a higher contribution score is assigned.

[0063] S300 constructs a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, identifies abrupt change nodes in the trajectory map, and determines whether the behavioral change constitutes a signal of credit risk shift based on the abrupt change nodes.

[0064] In this step, a multi-dimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. This is to achieve dynamic monitoring of user credit behavior. The trajectory map is user-centric, vertically recording the user's time-series performance across key dimensions. Each dimension forms a behavior curve, with time on the horizontal axis and standardized scores or behavioral indicator values ​​on the vertical axis. As data continues to be input, the trajectory map is continuously updated, clearly reflecting the evolution trend and stability of user behavior. Through sliding window and statistical anomaly detection methods, the system continuously monitors the curves of each dimension, capturing abrupt change nodes in the behavior curves—that is, abnormal jumps or interruptions in behavioral indicators at a certain point in time.

[0065] After identifying abrupt change events, the system further combines historical behavioral baselines, dimensional importance, and anomaly persistence to determine risk shifts. If the abrupt change has not occurred historically, the magnitude of the change exceeds a threshold, and it occurs on a highly sensitive dimension, it is identified as a credit risk shift signal, triggering a scoring correction process. Once identified, it is marked as a high-risk shift signal, and its credit score is lowered.

[0066] S400 quantifies the positive or negative impact of each behavioral characteristic on the final score, obtains updated evidence information based on the scoring elements, and derives new suggested scores based on the evidence information.

[0067] In this step, the positive or negative impact of each behavioral feature on the final score is quantified based on the final score. To address the issues of unexplainable scoring results and the inability of users to provide proactive feedback, the causal decomposition of each behavioral feature constituting the score is performed based on the final score. Marginal impact analysis is conducted on each feature using local interpretable models such as SHAP value or LIME to quantify its positive or negative impact on the final score.

[0068] After the scoring results are explained, if a user disagrees with a specific scoring factor, they can submit updated supporting evidence through the front-end interface, such as payment receipts, dispute withdrawal notices, updated professional certificates, or successful platform appeal records. Through OCR recognition and text semantic analysis, the user-provided evidence is matched and correlated with the original scoring elements, automatically filtering items with corrective value and replacing or correcting relevant feature values. After obtaining valid materials, the model re-runs the scoring process, generating new suggested scores and comparing them with the original scores.

[0069] like Figure 2As shown, in a preferred embodiment of the present invention, the steps of constructing edge weights to form a multidimensional credit graph based on proxy variables, identifying similar credit behavior communities based on existing data, obtaining similar user groups, and determining contribution scores based on the contribution of new users specifically include:

[0070] S201. Determine the "white account" user based on credit data. The "white account" user is a user with no credit history. Obtain proxy variables based on the "white account" user. The proxy variables include public utility payment records such as water, electricity and gas, and the fulfillment rate and return rate on e-commerce platforms.

[0071] In this step, credit data is used to identify users with no credit history. During credit data processing, the user's historical behavior information from banks, consumer finance companies, and credit platforms is analyzed to determine whether they have a complete credit record. If a user has never opened a credit card, has no loan record, and has no repayment history, they can be identified as a user with no credit history. Traditional scoring models lack core assessment criteria for users with no credit history; therefore, it is necessary to construct proxy variables using alternative data to supplement their credit profile.

[0072] The proxy variables mainly cover two aspects: First, public utility payment records such as water, electricity, and gas, including the continuity of payments, whether there are any interruptions, and fluctuations in bill amounts. For example, if a user pays their water, electricity, and gas bills on time for 12 consecutive months, it indicates good residential stability and payment ability, which can positively reflect their willingness to fulfill their credit responsibilities. Second, fulfillment behavior data on e-commerce platforms, including order completion rate, return rate, frequency of negative reviews, and after-sales dispute records. For example, if a user completes 50 e-commerce transactions in the past 6 months, returns only 2 items, and provides positive reviews, it indicates stable transaction behavior and a strong willingness to fulfill obligations, which can be regarded as a positive credit proxy indicator.

[0073] S202: Determine financial behavior capability based on proxy variables, construct edge weights based on proxy variables to form a multi-dimensional credit graph, obtain existing data, identify similar credit behavior communities based on existing data, and obtain the location of similar people.

[0074] In this step, financial capacity is determined based on proxy variables. To assess a user's financial capacity in the absence of credit, a quantitative model is first constructed based on proxy variables to evaluate their sense of responsibility, willingness to fulfill obligations, and stability of life. Proxy variables include public utility payment records (such as the continuity and timeliness of utility bills), e-commerce fulfillment performance (such as order completion rate, return rate, and review content), communication payment habits, and device location stability. By setting weights and thresholds for each variable, it is possible to preliminarily infer whether a user possesses the financial capacity to fulfill obligations on time and make continuous payments. For example, a user who has made continuous payments for 12 months, has an e-commerce return rate of less than 5%, and has active and stable phone activity can be judged to have strong financial capacity.

[0075] Based on the aforementioned proxy variables, a multidimensional feature vector is constructed. A similarity metric (cosine similarity) is used to calculate the behavioral similarity between any two users. Users are represented as nodes in a graph structure, connected by edges whose weights reflect the degree of behavioral similarity, thus forming a multidimensional credit graph. After the graph is built, a graph embedding algorithm is used to map the nodes to a low-dimensional space, and a clustering method (K-means) is combined to identify user groups with homogeneous features, i.e., communities with similar credit behavior. For example, if most users in a community share the common characteristics of stable payments, good e-commerce performance, but no credit history, it can be inferred that the overall risk of this community is relatively low.

[0076] S203, obtain the contribution of existing users with no credit history, the contribution includes continuous payment, positive reviews on e-commerce platforms, and unstable location, and determine the contribution score based on the contribution of users with no credit history.

[0077] In this step, we obtain the contribution score of existing users without credit history. To assess the matching quality and behavioral credibility of these users in the credit graph, we calculate their performance on key proxy variable dimensions to obtain a contribution score. The contribution score measures the consistency between the behavior of users without credit history and the characteristics of their similar community centers. It mainly includes three aspects: First, continuous payment, i.e., whether the user has consistently paid their utility bills (water, electricity, gas, phone bills, etc.). If the user's payments have been uninterrupted and have low volatility over the past 12 months, their payment behavior can be considered stable, with a high positive contribution. Second, positive reviews on e-commerce platforms, referring to the user's order completion rate, positive review rate, and low dispute rate on e-commerce platforms. If the positive review rate is above 90%, it indicates strong fulfillment ability and stable consumption behavior, constituting a clear positive credit signal. Third, location instability, reflecting whether the user's mobile device's geographical location frequently changes across regions or jumps in a short period. Frequent changes of residence or abnormal whereabouts are considered negative contribution factors, potentially indicating unstable living conditions or hidden behavior.

[0078] When constructing the contribution score, the aforementioned proxy variables are standardized and assigned feature weights. A matching score is calculated based on their fit to the target community's central feature vector. For example, if a user's payment continuity and positive review rate are highly similar to the target group, but their location volatility is slightly higher, a contribution score of 0.76 can be output, indicating that their overall behavior is reliable but has a slight bias. Ultimately, this contribution score is not only used to determine the credibility of their affiliation within the current community but also serves as one of the important reference indicators for their credit score weighting.

[0079] like Figure 3 As shown, in a preferred embodiment of the present invention, the step of constructing a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, identifying abrupt change nodes in the trajectory map, and determining whether the behavioral change constitutes a signal of credit risk shift based on the abrupt change nodes specifically includes:

[0080] S301, Obtain real-time behavioral data streams, including financial, e-commerce, and public service data, and obtain platform data by connecting to the platform through the authorized API interface.

[0081] In this step, real-time behavioral data streams are acquired to enable dynamic identification and continuous updating of user credit status. A behavioral evolution sequence is constructed using these data streams, which primarily include three categories of information: financial, e-commerce, and public services. Financial data covers account balance changes, repayment behavior, credit usage, and overdue warnings. For example, a user's failure to make an automatic repayment at a bank triggers a negative behavior node. E-commerce data includes order completion status, review records, return frequency, and shopping cycles; for instance, three consecutive refund requests are considered a potential performance deviation signal. Public service data refers to the timeliness and amount fluctuations in utility bill payments (water, electricity, gas), as well as any sudden changes in urban commuting patterns. For example, a user's phone bill being suspended for two consecutive months or frequent cross-province location signal changes may indicate risk.

[0082] All data acquisition relies on standardized API interfaces provided by various platforms after user authorization. By establishing interface connections with banks, e-commerce platforms, and public utility service platforms, credit-related behavioral data is automatically retrieved with user authorization and integrated into the user's credit trajectory according to timestamps. This process enables high-frequency collection and real-time synchronization of key credit behaviors, ensuring that the scoring model is always dynamically updated based on the latest and most accurate behavioral status.

[0083] S302, Construct a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data. The dimensions of the trajectory map represent key credit factors, and identify abrupt change nodes in the trajectory map.

[0084] In this step, a multi-dimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. Based on real-time behavioral data streams obtained from financial, e-commerce, and public service platforms, a user-centric multi-dimensional credit behavior trajectory map is built. This trajectory map uses time as the horizontal axis and multiple key credit factors as the vertical dimensions. Each factor dimension corresponds to a time-updated behavioral sequence curve, reflecting the changing trends of a user's credit behavior across different dimensions. Commonly used dimensions include repayment records, e-commerce fulfillment status, payment behavior, and location stability. For example, if a user's utility payment continuity dimension has remained stable over the past six months, but two consecutive overdue payments occur in the current month, this will appear as a significant abrupt change in that dimension's trajectory.

[0085] To identify abrupt change nodes in the trajectory graph, statistical anomaly detection methods such as Z-score, sliding window mean deviation, or local outlier factor (LOF) are used to dynamically monitor each behavioral curve. When the behavioral value at a certain point in time significantly deviates from its historical average level and crosses a set threshold, it can be marked as an abrupt change node. For example, if a user's average e-commerce return rate over the past 12 months is 2%, but suddenly rises to 12% this month, this node will be identified as an abrupt change and used as an early signal of potential fulfillment risk in score correction.

[0086] S303: Based on the mutation node, determine whether the behavioral change constitutes a credit risk shift signal. If a credit risk shift is determined, adjust the score downward by calculating the correction factor and retain the node record for trend monitoring.

[0087] In this step, the abrupt change node is used to determine whether the behavioral change constitutes a signal of credit risk shift. After identifying the abrupt change node in the multidimensional credit behavior trajectory map, it is necessary to further determine whether the behavioral change constitutes a substantial credit risk shift signal. The judgment process comprehensively considers factors such as the magnitude of the change, its duration, the degree of deviation from the historical baseline, and the credit sensitivity of the dimension to which it belongs. If the change occurs in a high-weight dimension, such as repayment behavior or e-commerce fulfillment, and the magnitude of the change exceeds the statistical threshold, it indicates that the behavior has significant abnormal characteristics. Taking a user's repayment behavior over the past 12 months as an example, if a user has always made timely repayments but has delayed twice this month, with the delay exceeding 10 days, this behavioral node is marked as abnormal, and its deviation relative to the historical baseline is further calculated. A correction factor is constructed by combining the importance weights of the behavioral dimensions.

[0088] like Figure 4 As shown, in a preferred embodiment of the present invention, the steps of quantifying the positive or negative impact of each behavioral feature on the final score based on the final score, obtaining updated evidence information based on the scoring elements, and obtaining a new suggested score value based on the evidence information specifically include:

[0089] S401, Obtain the corrected final score, quantify the positive or negative impact of each behavioral feature on the final score based on the final score, and visualize the quantification results.

[0090] In this step, the corrected final score is obtained. After identifying mutation nodes and calculating correction factors, the original score and correction factors are superimposed to obtain the corrected final score. This final score not only reflects the user's current overall credit status but also includes dynamic adjustments for short-term risk biases. To enhance the transparency and interpretability of the scoring results, the influence of each behavioral feature constituting the score is further quantified. Algorithms such as SHAP (Shapley Additive Explanations) are used to perform inversion analysis on the scoring model, calculating the marginal contribution of each feature in the model's prediction results. Each behavioral factor is assigned a positive or negative scoring weight, representing its effect on improving or deducting the final score. For example, if a user's final score is 82 points, with continuous utility payment contributing +10 points, occupational volatility contributing -6 points, and e-commerce return rate contributing -4 points, the system stores these quantified results in the form of scores, forming the scoring interpretation structure.

[0091] To enhance user understanding of the scoring structure, all feature contributions are visualized, often using bar charts, score composition graphs, or hierarchical score distribution charts. The charts indicate the direction and intensity of each factor's influence on the score. For example, a chart might show payment behavior and e-commerce fulfillment in the positive area, while unstable device location is in the negative area, helping users quickly understand the primary causes of their scores and identify potential areas for behavioral optimization.

[0092] S402, Obtain the scoring components, which include credit enhancement factors, credit risk factors, and neutral characteristics. Obtain updated evidence information based on the scoring components. The evidence information is used for scoring appeals.

[0093] In this step, the scoring components are obtained. While outputting the final score, these components are further extracted, categorizing all behavioral characteristics involved in the calculation into three types: credit enhancement factors, credit risk factors, and neutral characteristics. Credit enhancement factors are behavioral characteristics that significantly positively impact the score, such as consistently paying utility bills, high online customer satisfaction rates, and stable employment records. Credit risk factors are characteristics that negatively impact the score, such as frequent credit card delinquencies, frequent returns, and frequent changes in location. Neutral characteristics are variables with a weak or insignificant impact on the score, such as occasional small refunds or minor payment delays. The scoring results are presented in a structured format, along with the direction and weight of each factor's influence, for user understanding and appeal purposes.

[0094] When users disagree with their rating, they can focus on specific risk factors or key deduction items based on the above classification results and submit corresponding updated supporting evidence. This evidence may include payment receipts for mischarged phone bills, records of dispute cancellations on e-commerce platforms, proof of employment, or evidence of address stability. The system uses OCR and semantic analysis to identify the content of the materials and match them with corresponding rating elements. After model validation, a partial re-rating process is triggered. For example, if a user's rating is deducted due to overdue phone bills, but they provide valid payment screenshots and operator bills, and the model comparison confirms a mischarge, the system will automatically adjust the value of that rating factor and update the rating result.

[0095] S403. Based on the evidence materials, a new suggested score is obtained. If the suggested score differs significantly from the original score and the evidence materials are credible, the score is corrected and the score correction path is recorded.

[0096] In this step, a new suggested score is obtained based on the evidence materials. When a user submits evidence materials related to the scoring result, the system verifies and updates the corresponding scoring factors based on these materials, then reruns the scoring model to generate a new suggested score. The material verification process first extracts the content of the voucher through OCR text recognition, and then confirms its correlation with the specific scoring factors through semantic matching and field comparison algorithms. For example, if a user is penalized for overdue phone bills, and the uploaded phone bill payment record is recognized as automatically paid and the payment time is exactly the same as the billing period, the model considers the material valid and corrects the value of the scoring factor from overdue to on-time payment. After updating the factors, the scoring model recalculates the results. If the generated new suggested score differs significantly from the original score, such as exceeding a set threshold of 5 points, and the evidence materials score well in the credibility assessment, then the suggested score will be adopted.

[0097] After the correction is completed, the scoring process will automatically record the complete path of this scoring adjustment, including the triggering factor, original value, modified value, material verification status, and the difference before and after the scoring, and generate a scoring correction log. Taking a user as an example, the initial score is 76 points, of which 6 points were mistakenly deducted due to a system delay in updating an e-commerce dispute record; after the user submits a revocation certificate and the model confirms its validity, the factor is corrected from "dispute exists" to "dispute cleared", and the new score is updated to 81 points. The system will archive this change as a scoring correction event in the user's credit history.

[0098] like Figure 5 As shown, an embodiment of the present invention provides a credit information data management and analysis system, the system comprising:

[0099] The credit information module 100 is used to acquire credit data information, which is obtained from multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators.

[0100] In this system, the credit information module 100 acquires credit data. This acquisition employs a multi-source heterogeneous data fusion mechanism, covering multiple sources including banks, consumer finance institutions, and telecommunications operators, to improve the completeness and representativeness of user credit profiles. At the bank level, it can access account transaction data, credit card usage records, repayment history, and loan contract information from traditional commercial banks and internet banks, paying particular attention to quantifiable indicators such as repayment frequency, bill amount ratio, and delinquency behavior. For example, if a user has multiple small loans in the past 12 months, without delinquency, but frequently switches loan platforms, this behavior will be used as a risk boundary factor in the model.

[0101] In the consumer finance dimension, by accessing data interfaces of consumer installment platforms, micro-loan systems and e-commerce financial products, we can extract users' performance in installment shopping, cash installment, BNPL (buy now, pay later), approval rate, credit limit utilization rate and dispute records to establish a model of users' willingness to assume responsibility and financial behavior habits.

[0102] Telecommunications operator data serves as an auxiliary dimension for credit scoring, providing users with detailed call records, number usage duration, payment cycle patterns, roaming records, and location trajectories to assess their life stability and credit consistency. For example, long-term arrears in phone bills, frequent changes in mobile phone numbers, or sudden changes in call patterns will be treated as potential credit default warning factors.

[0103] The contribution scoring module 200 is used to construct edge weights based on proxy variables to form a multi-dimensional credit graph, identify similar credit behavior communities based on existing data, obtain similar groups of people, and determine contribution scores based on the contribution of new users.

[0104] In this system, the contribution scoring module 200 constructs edge weights based on proxy variables to form a multidimensional credit graph. To model the credit behavior of users without credit history, edge weights are first constructed based on proxy variables to form a multidimensional credit graph reflecting the relationships between user behavioral characteristics. Specifically, each user constructs a feature vector through their proxy variables, such as the continuity of utility bill payments, e-commerce fulfillment rate, device location stability, and phone bill payment habits. The graph structure constructed in this way not only preserves the behavioral similarity relationships between users but also maps the implicit behavioral trend proximity in the graph.

[0105] Next, blank users are integrated into the graph. Based on the edge weights between their proxy feature vectors and labeled users, the most similar Top-K users or communities are identified. Then, based on the historical average and fluctuations of users within the community, the fitting credit behavior trends of blank users across various dimensions are preliminarily estimated. The degree of fitting and perturbation of blank users to the feature centers of adjacent communities in the graph structure is evaluated. If their behavior is highly consistent with the core features of the community and does not cause significant structural perturbation, a higher contribution score is assigned.

[0106] The risk identification module 300 is used to construct a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, identify abrupt change nodes in the trajectory map, and determine whether the behavioral change constitutes a signal of credit risk shift based on the abrupt change nodes.

[0107] In this system, the risk identification module 300 constructs a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data. This trajectory map, centered on the user, vertically records their time-series performance across key dimensions. Each dimension forms a behavior curve, with time on the horizontal axis and standardized scores or behavioral indicator values ​​on the vertical axis. As data continues to be input, the trajectory map is constantly updated, clearly reflecting the evolution and stability of user behavior. Through sliding window and statistical anomaly detection methods, the system continuously monitors the curves of each dimension, capturing abrupt change nodes in the behavior curves—that is, abnormal jumps or interruptions in behavioral indicators at a certain point in time.

[0108] After identifying abrupt change events, the system further combines historical behavioral baselines, dimensional importance, and anomaly persistence to determine risk shifts. If the abrupt change has not occurred historically, the magnitude of the change exceeds a threshold, and it occurs on a highly sensitive dimension, it is identified as a credit risk shift signal, triggering a scoring correction process. Once identified, it is marked as a high-risk shift signal, and its credit score is lowered.

[0109] The scoring interpretation module 400 is used to quantify the positive or negative impact of each behavioral feature on the final score, obtain updated evidence information based on the scoring elements, and obtain new suggested scores based on the evidence information.

[0110] In this system, the rating interpretation module 400 quantifies the positive or negative impact of each behavioral feature on the final rating. To address the issues of uninterpretable rating results and the inability of users to provide proactive feedback, the module decomposes the causal relationships of each behavioral feature that constitutes the rating based on the final rating result. It then performs marginal impact analysis on each feature using local interpretable models such as SHAP value or LIME to quantify its positive or negative impact on the final rating.

[0111] After the scoring results are explained, if a user disagrees with a specific scoring factor, they can submit updated supporting evidence through the front-end interface, such as payment receipts, dispute withdrawal notices, updated professional certificates, or successful platform appeal records. Through OCR recognition and text semantic analysis, the user-provided evidence is matched and correlated with the original scoring elements, automatically filtering items with corrective value and replacing or correcting relevant feature values. After obtaining valid materials, the model re-runs the scoring process, generating new suggested scores and comparing them with the original scores.

[0112] like Figure 6 As shown, in a preferred embodiment of the present invention, the contribution scoring module 200 includes:

[0113] The proxy variable unit 201 is used to determine the blank user based on credit data. The blank user is a user with no credit history. Proxy variables are obtained based on the blank user. The proxy variables include public utility payment records such as water, electricity and gas, and the fulfillment rate and return rate in e-commerce platforms.

[0114] In this module, proxy variable unit 201 identifies users with no credit history based on credit data. During credit data processing, it analyzes users' historical behavior information from banks, consumer finance companies, and credit platforms to determine whether they have a complete credit record. If a user has never opened a credit card, has no loan record, or has no repayment history, they can be identified as a user with no credit history. Traditional scoring models lack core assessment criteria for users with no credit history; therefore, proxy variables need to be constructed using alternative data to supplement their credit profile.

[0115] The proxy variables mainly cover two aspects: First, public utility payment records such as water, electricity, and gas, including the continuity of payments, whether there are any interruptions, and fluctuations in bill amounts. For example, if a user pays their water, electricity, and gas bills on time for 12 consecutive months, it indicates good residential stability and payment ability, which can positively reflect their willingness to fulfill their credit responsibilities. Second, fulfillment behavior data on e-commerce platforms, including order completion rate, return rate, frequency of negative reviews, and after-sales dispute records. For example, if a user completes 50 e-commerce transactions in the past 6 months, returns only 2 items, and provides positive reviews, it indicates stable transaction behavior and a strong willingness to fulfill obligations, which can be regarded as a positive credit proxy indicator.

[0116] Credit graph unit 202 is used to determine financial behavior capability based on proxy variables, construct edge weights based on proxy variables to form a multi-dimensional credit graph, obtain existing data, identify similar credit behavior communities based on existing data, and obtain the location of similar people.

[0117] In this module, the Credit Graph Unit 202 determines financial behavior capacity based on proxy variables. To assess a user's financial behavior capacity in the absence of credit, it first quantifies and models their sense of responsibility, willingness to fulfill obligations, and life stability based on proxy variables. Proxy variables include public utility payment records (such as the continuity and timeliness of utility bills), e-commerce performance (such as order completion rate, return rate, and review content), communication payment habits, and device location stability. By setting weights and thresholds for each variable, it is possible to preliminarily infer whether a user possesses the financial behavior capacity to fulfill obligations on time and make continuous payments. For example, a user who has made continuous payments for 12 months, has an e-commerce return rate of less than 5%, and has active and stable phone calls can be judged to have strong financial behavior capacity.

[0118] Based on the aforementioned proxy variables, a multidimensional feature vector is constructed. A similarity metric (cosine similarity) is used to calculate the behavioral similarity between any two users. Users are represented as nodes in a graph structure, connected by edges whose weights reflect the degree of behavioral similarity, thus forming a multidimensional credit graph. After the graph is built, a graph embedding algorithm is used to map the nodes to a low-dimensional space, and a clustering method (K-means) is combined to identify user groups with homogeneous features, i.e., communities with similar credit behavior. For example, if most users in a community share the common characteristics of stable payments, good e-commerce performance, but no credit history, it can be inferred that the overall risk of this community is relatively low.

[0119] The contribution scoring unit 203 is used to obtain the contribution of existing new users. The contribution includes continuous payment, positive reviews on e-commerce platforms, and unstable location. The contribution score is determined based on the contribution of new users.

[0120] In this module, the contribution scoring unit 203 acquires the contribution scores of existing "no-show" users. To assess the matching quality and behavioral credibility of these users in the credit graph, a contribution score is obtained by calculating their performance on key proxy variable dimensions. Contribution score measures the consistency between the behavior of "no-show" users and the characteristics of their affiliated similar community centers. It mainly includes three aspects: First, continuous payment, i.e., whether the user has consistently paid their utility bills (water, electricity, gas, phone bills, etc.). If the user's payments have been uninterrupted and have low volatility over the past 12 months, their payment behavior can be considered stable, indicating a high positive contribution. Second, positive reviews on e-commerce platforms, referring to the user's order completion rate, positive review rate, and low dispute rate on e-commerce platforms. If the positive review rate is above 90%, it indicates strong fulfillment ability and stable consumption behavior, constituting a clear positive credit signal. Third, location instability, reflecting whether the user's mobile device's geographical location frequently changes across regions or jumps in a short period. Frequent changes of residence or abnormal whereabouts are considered negative contribution factors, potentially indicating unstable living conditions or hidden behavior.

[0121] When constructing the contribution score, the aforementioned proxy variables are standardized and assigned feature weights. A matching score is calculated based on their fit to the target community's central feature vector. For example, if a user's payment continuity and positive review rate are highly similar to the target group, but their location volatility is slightly higher, a contribution score of 0.76 can be output, indicating that their overall behavior is reliable but has a slight bias. Ultimately, this contribution score is not only used to determine the credibility of their affiliation within the current community but also serves as one of the important reference indicators for their credit score weighting.

[0122] like Figure 7 As shown, in a preferred embodiment of the present invention, the risk identification module 300 includes:

[0123] The real-time behavior unit 301 is used to acquire real-time behavior data streams, which include financial, e-commerce and public service data, and to obtain platform data by connecting to the platform through authorized API interfaces.

[0124] In this module, the real-time behavior unit 301 acquires real-time behavior data streams. To achieve dynamic identification and continuous updating of users' credit status, a behavior evolution sequence is constructed by acquiring these data streams. This data stream mainly includes three categories of information: financial, e-commerce, and public services. The financial data section covers account balance changes, repayment behavior, credit usage, and overdue warnings. For example, a user's failure to make an automatic repayment at a bank triggers a negative behavior node. E-commerce data includes order completion status, review records, return frequency, and shopping cycles. For instance, three consecutive after-sales refunds are considered a potential performance deviation signal. Public service data refers to the timeliness and amount fluctuations of utility bill payments (water, electricity, gas), as well as any sudden changes in urban commuting patterns. For example, a user's phone bill being suspended for two consecutive months or frequent cross-province location signals from their device may indicate risk.

[0125] All data acquisition relies on standardized API interfaces provided by various platforms after user authorization. By establishing interface connections with banks, e-commerce platforms, and public utility service platforms, credit-related behavioral data is automatically retrieved with user authorization and integrated into the user's credit trajectory according to timestamps. This process enables high-frequency collection and real-time synchronization of key credit behaviors, ensuring that the scoring model is always dynamically updated based on the latest and most accurate behavioral status.

[0126] The behavior trajectory graph unit 302 is used to construct a multi-dimensional credit behavior trajectory graph based on real-time behavior data streams and platform data. The trajectory graph dimensions represent key credit factors and identify abrupt change nodes in the trajectory graph.

[0127] In this module, the behavior trajectory graph unit 302 constructs a multi-dimensional credit behavior trajectory graph based on real-time behavior data streams and platform data. Based on real-time behavior data streams obtained from financial, e-commerce, and public service platforms, it builds a user-centric multi-dimensional credit behavior trajectory graph. This trajectory graph uses time as the horizontal axis and multiple key credit factors as the vertical dimensions. Each factor dimension corresponds to a time-updated behavior sequence curve, reflecting the changing trends of user credit behavior across different dimensions. Commonly used dimensions include repayment records, e-commerce fulfillment status, payment behavior, and location stability. For example, if a user's utility payment continuity dimension has remained stable over the past six months, but two consecutive overdue payments occur in the current month, this will manifest as a significant abrupt change in that dimension's trajectory.

[0128] To identify abrupt change nodes in the trajectory graph, statistical anomaly detection methods such as Z-score, sliding window mean deviation, or local outlier factor (LOF) are used to dynamically monitor each behavioral curve. When the behavioral value at a certain point in time significantly deviates from its historical average level and crosses a set threshold, it can be marked as an abrupt change node. For example, if a user's average e-commerce return rate over the past 12 months is 2%, but suddenly rises to 12% this month, this node will be identified as an abrupt change and used as an early signal of potential fulfillment risk in score correction.

[0129] Risk offset unit 303 is used to determine whether the change in behavior constitutes a credit risk offset signal based on the mutation node. If a credit risk offset is determined to exist, the score is downgraded by calculating the correction factor, and the node record is retained for trend monitoring.

[0130] In this module, the risk offset unit 303 determines whether a behavioral change constitutes a credit risk offset signal based on abrupt change nodes. After identifying abrupt change nodes in the multidimensional credit behavior trajectory map, it is necessary to further determine whether the behavioral change constitutes a substantial credit risk offset signal. The judgment process comprehensively considers factors such as the magnitude of the change, its duration, the degree of deviation from the historical baseline, and the credit sensitivity of the dimension to which it belongs. If the change occurs in a high-weight dimension, such as repayment behavior or e-commerce fulfillment, and the magnitude of the change exceeds the statistical threshold, it indicates that the behavior has significant abnormal characteristics. Taking a user's repayment behavior over the past 12 months as an example, if a user has consistently made timely repayments but has delayed twice this month for more than 10 days, this behavioral node is marked as abnormal, and its offset relative to the historical baseline is further calculated. A correction factor is constructed by combining the importance weights of the behavioral dimensions.

[0131] like Figure 8 As shown, in a preferred embodiment of the present invention, the scoring interpretation module 400 includes:

[0132] The scoring interpretation unit 401 is used to obtain the corrected final score, quantify the positive or negative impact of each behavioral feature on the final score based on the final score, and visualize the quantification results.

[0133] In this module, the scoring interpretation unit 401 obtains the corrected final score. After completing the identification of mutation nodes and the calculation of correction factors, the original score and the correction factors are superimposed to obtain the corrected final score. This final score not only reflects the user's current comprehensive credit status but also includes the dynamic adjustment results for short-term risk bias. To enhance the transparency and interpretability of the scoring results, the influence of each behavioral feature constituting the score is further quantified. Algorithms such as SHAP (Shapley Additive Explanations) are used to perform inversion analysis on the scoring model, calculating the marginal contribution value of each feature in the model's prediction results. Each behavioral factor is assigned a positive or negative scoring weight, indicating its effect on improving or deducting the final score. For example, if a user's final score is 82 points, with continuous utility payment contributing +10 points, occupational volatility contributing -6 points, and e-commerce return rate contributing -4 points, the system stores these quantified results in the form of scores, forming the scoring interpretation structure.

[0134] To enhance user understanding of the scoring structure, all feature contributions are visualized, often using bar charts, score composition graphs, or hierarchical score distribution charts. The charts indicate the direction and intensity of each factor's influence on the score. For example, a chart might show payment behavior and e-commerce fulfillment in the positive area, while unstable device location is in the negative area, helping users quickly understand the primary causes of their scores and identify potential areas for behavioral optimization.

[0135] The material information update unit 402 is used to obtain the scoring components, which include credit enhancement factors, credit risk factors and neutral characteristics. Based on the scoring components, it obtains updated evidence material information, which is used for scoring appeals.

[0136] In this module, the material information update unit 402 acquires the scoring components. While outputting the final score, it further extracts these components, classifying all behavioral characteristics involved in the calculation into three types: credit enhancement factors, credit risk factors, and neutral features. Credit enhancement factors are behavioral characteristics that significantly positively impact the score, such as consistently paying utility bills, high online customer satisfaction rates, and stable employment records. Credit risk factors are characteristics that negatively impact the score, such as frequent credit card delinquencies, frequent returns, and frequent changes in location. Neutral features are variables with a weak or insignificant impact on the score, such as occasional small refunds or minor payment delays. The scoring results are presented in a structured format, along with the direction and weight of each factor's influence, for user understanding and appeal purposes.

[0137] When users disagree with their rating, they can focus on specific risk factors or key deduction items based on the above classification results and submit corresponding updated supporting evidence. This evidence may include payment receipts for mischarged phone bills, records of dispute cancellations on e-commerce platforms, proof of employment, or evidence of address stability. The system uses OCR and semantic analysis to identify the content of the materials and match them with corresponding rating elements. After model validation, a partial re-rating process is triggered. For example, if a user's rating is deducted due to overdue phone bills, but they provide valid payment screenshots and operator bills, and the model comparison confirms a mischarge, the system will automatically adjust the value of that rating factor and update the rating result.

[0138] The scoring correction unit 403 is used to obtain a new scoring suggestion value based on the evidence material information. If the suggested value is significantly different from the original score and the evidence material information is credible, the score is corrected and the scoring correction path is recorded.

[0139] In this module, the scoring correction unit 403 obtains a new scoring suggestion value based on the evidence material information. When a user submits evidence material information related to the scoring result, the system will verify and update the corresponding scoring factors based on the material, and then rerun the scoring model to generate a new scoring suggestion value. The material verification process first extracts the content of the voucher through OCR text recognition, and then confirms its correlation with the specific scoring factor through semantic matching and field comparison algorithms. For example, if a user is penalized for overdue phone bills, and the uploaded phone bill payment record is recognized as automatically paid and the payment time is completely consistent with the billing period, the model considers the material valid and corrects the value of the scoring factor from overdue to on-time payment. After updating the factors, the scoring model recalculates the results. If the generated new scoring suggestion value differs significantly from the original score, such as exceeding the set threshold of 5 points, and the evidence material scores well in the credibility assessment, then the scoring suggestion will be adopted.

[0140] After the correction is completed, the scoring process will automatically record the complete path of this scoring adjustment, including the triggering factor, original value, modified value, material verification status, and the difference before and after the scoring, and generate a scoring correction log. Taking a user as an example, the initial score is 76 points, of which 6 points were mistakenly deducted due to a system delay in updating an e-commerce dispute record; after the user submits a revocation certificate and the model confirms its validity, the factor is corrected from "dispute exists" to "dispute cleared", and the new score is updated to 81 points. The system will archive this change as a scoring correction event in the user's credit history.

[0141] In one embodiment, a computer device is provided, the computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, performs the following steps:

[0142] Credit information is obtained through multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators.

[0143] A multidimensional credit graph is formed by constructing edge weights based on proxy variables. Similar credit behavior communities are identified based on existing data to obtain similar groups of people. Contribution scores are determined based on the contribution of new users.

[0144] A multidimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. Abrupt nodes in the trajectory map are identified, and it is determined whether the behavioral change constitutes a signal of credit risk shift based on the abrupt nodes.

[0145] The final score quantifies the positive or negative impact of each behavioral characteristic on the final score, updates the evidence materials based on the scoring elements, and then derives new suggested scores based on the evidence materials.

[0146] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, causes the processor to perform the following steps:

[0147] Credit information is obtained through multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators.

[0148] A multidimensional credit graph is formed by constructing edge weights based on proxy variables. Similar credit behavior communities are identified based on existing data to obtain similar groups of people. Contribution scores are determined based on the contribution of new users.

[0149] A multidimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. Abrupt nodes in the trajectory map are identified, and it is determined whether the behavioral change constitutes a signal of credit risk shift based on the abrupt nodes.

[0150] The final score quantifies the positive or negative impact of each behavioral characteristic on the final score, updates the evidence materials based on the scoring elements, and then derives new suggested scores based on the evidence materials.

[0151] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0152] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0153] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0154] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0155] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing and analyzing credit information data, characterized in that, The method includes: Credit information is obtained through multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators. A multidimensional credit graph is formed by constructing edge weights based on proxy variables. Target communities with similar credit behaviors are identified based on existing data. Contribution scores are determined based on the contribution of new users. Contribution score measures the degree of consistency between the behavior of new users and the central characteristics of the target community to which they belong. The contribution score is used as a reference indicator for the weighting of credit scores. A multidimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. Abrupt change nodes are identified within the trajectory map, and based on these nodes, it is determined whether the behavioral change constitutes a signal of credit risk shift. Specifically, this includes: Acquire real-time behavioral data streams, including financial, e-commerce, and public service data, and obtain platform data by connecting to the platform through authorized API interfaces; A multidimensional credit behavior trajectory map is constructed based on real-time behavioral data streams and platform data. The dimensions of the trajectory map represent key credit factors, and abrupt change nodes are identified in the trajectory map. Based on the mutation node, it is determined whether the change in behavior constitutes a signal of credit risk deviation. If a credit risk deviation is determined, the credit score is adjusted down by calculating the correction factor to obtain the final score, and the node record is retained for trend monitoring. The positive or negative impact of each behavioral characteristic on the final score is quantified based on the final score. Updated evidence information is obtained based on the scoring elements, and new suggested scores are derived based on the evidence information. The steps of constructing a multidimensional credit graph based on proxy variables and edge weights, identifying similar credit behavior communities based on existing data, locating similar user groups, and determining contribution scores based on the contribution of new users specifically include: The system identifies "white-listed" users based on credit data. These white-listed users are those with no credit history. The system also obtains proxy variables based on these white-listed users, including utility bill payment records and fulfillment and return rates on e-commerce platforms. Financial behavior capability is determined based on proxy variables, and a multi-dimensional credit graph is formed by constructing edge weights based on proxy variables. Existing data is obtained, and communities with similar credit behavior are identified based on the existing data to obtain the location of similar people. Obtain the contribution score of existing new users, including continuous payment, positive e-commerce reviews, and unstable location. Determine the contribution score based on the contribution score of new users. The steps of quantifying the positive or negative impact of each behavioral feature on the final score based on the final score, obtaining updated evidence information based on the scoring elements, and obtaining new suggested scores based on the evidence information specifically include: Obtain the final score, quantify the positive or negative impact of each behavioral feature on the final score, and visualize the quantification results. Obtain the scoring components, which include credit enhancement factors, credit risk factors, and neutral characteristics. Based on the scoring components, obtain updated evidentiary information, which is used for scoring appeals. If a new suggested score is obtained based on the evidence, and the suggested score differs significantly from the original score and the evidence is credible, the score is corrected and the score correction path is recorded.

2. The method for managing and analyzing credit information data according to claim 1, characterized in that, The platforms include bank credit reporting, payment platforms, e-commerce platforms, and operators.

3. A credit information data management and analysis system, characterized in that, The system includes: The credit information module acquires credit data information, which is obtained from multiple heterogeneous sources, including data from banks, consumer finance companies, and telecom operators. The contribution scoring module constructs edge weights based on proxy variables to form a multidimensional credit graph. It identifies communities with similar credit behaviors based on existing data to obtain target communities. The contribution score is determined based on the contribution of new users. The contribution score measures the degree of consistency between the behavior of new users and the central characteristics of the target community to which they belong. The contribution score is used as a reference indicator for the weight of the credit score. The risk identification module constructs a multi-dimensional credit behavior trajectory map based on real-time behavioral data streams and platform data, identifies abrupt change nodes in the trajectory map, and determines whether the behavioral change constitutes a signal of credit risk shift based on the abrupt change nodes; the risk identification module includes: The real-time behavior unit acquires real-time behavior data streams, including financial, e-commerce, and public service data, and obtains platform data by connecting to the platform through authorized API interfaces. The behavioral trajectory graph unit constructs a multi-dimensional credit behavior trajectory graph based on real-time behavioral data streams and platform data. The trajectory graph dimensions represent key credit factors and identify abrupt change nodes in the trajectory graph. The risk offset unit determines whether the change in behavior constitutes a credit risk offset signal based on the mutation node. If a credit risk offset is determined to exist, the credit score is adjusted down by calculating the correction factor to obtain the final score, and the node record is retained for trend monitoring. The scoring interpretation module quantifies the positive or negative impact of each behavioral feature on the final score, obtains updated evidence information based on the scoring elements, and derives new suggested scores based on the evidence information. The contribution scoring module includes: The proxy variable unit determines blank users based on credit data. Blank users are users with no credit history. Proxy variables are obtained based on blank users. Proxy variables include utility bill payment records and fulfillment rate and return rate on e-commerce platforms. The credit graph unit determines financial behavior capability based on proxy variables, constructs edge weights based on proxy variables to form a multi-dimensional credit graph, acquires existing data, identifies similar credit behavior communities based on existing data, and obtains the location of similar people. The contribution score unit obtains the contribution score of existing new users, which includes continuous payment, positive e-commerce reviews, and unstable location. The contribution score is determined based on the contribution score of new users. The scoring interpretation module includes: The scoring interpretation unit obtains the final score, quantifies the positive or negative impact of each behavioral feature on the final score, and visualizes the quantification results. The material information update unit obtains the scoring components, which include credit enhancement factors, credit risk factors, and neutral characteristics. Based on the scoring components, it obtains updated evidence material information, which is used for scoring appeals. The scoring correction unit obtains a new suggested scoring value based on the evidence material information. If the suggested value differs significantly from the original scoring and the evidence material information is credible, the scoring is corrected and the scoring correction path is recorded.

4. The credit information data management and analysis system according to claim 3, characterized in that, The platforms include bank credit reporting, payment platforms, e-commerce platforms, and operators.

Images