Credit assessment-based detention-free rental service management method and system

Abstract

The invention discloses a detention-free lease service management method and system based on credit evaluation, and relates to the technical field of service management. The method comprises the following steps: traversing a user set in a detention-free lease service, and constructing a detention-free lease multi-source behavior data set; constructing a detention-free leasing business relation graph; constructing a credit homogamete graph and a risk heterogamete graph; performing feature analysis on the credit homogamete graph and the risk heterogamete graph to obtain user stable credit features and user risk anomaly features; performing feature fusion based on the user stable credit feature and the user risk anomaly feature to obtain a user comprehensive credit assessment result; and executing detention-free lease service management decision analysis on the target user to obtain a detention-free lease management strategy. The technical problem of insufficient service management decision reliability caused by inaccurate user credit evaluation in the detention-free leasing service in the prior art is solved, and the technical effects of improving the credit evaluation precision and enhancing the service management decision reliability through multi-source behavior data association analysis are achieved.

Description

Technical Field

[0001] This invention relates to the field of business management technology, specifically to a method and system for managing deposit-free leasing business based on credit assessment. Background Technology

[0002] With the rapid development of the sharing economy and asset-light operation models, deposit-free leasing services have been widely adopted in areas such as digital equipment leasing, transportation vehicle leasing, and office equipment leasing. While the deposit-free leasing model lowers the user entry barrier and improves business conversion rates and market coverage, it also places higher demands on risk control and credit assessment. Existing deposit-free leasing services typically rely on users' historical credit scores, payment records, or simple behavioral rules for credit judgment, with some systems incorporating third-party credit data for supplementary assessment. However, these assessment methods often focus on single-dimensional or static feature analysis, lacking structured modeling of the relationships between multi-source behavioral data, making it difficult to identify implicit connections between users, abnormal group behavior, or risk propagation paths. When complex risks such as address sharing and frequent abnormal device transfers exist, traditional credit assessment models often fail to effectively identify them, leading to inaccurate credit assessment results and consequently affecting the reliability of deposit-free leasing business management decisions. Summary of the Invention

[0003] This application provides a credit assessment-based management method and system for deposit-free leasing business, which solves the technical problem in the prior art where inaccurate user credit assessment leads to insufficient reliability of business management decisions.

[0004] The first aspect of this application provides a method for managing deposit-free leasing business based on credit assessment, the method comprising:

[0005] The system iterates through the user set in the deposit-free rental business, collecting user information, rental order information, equipment information, address information, and payment account information for each user to construct a multi-source behavioral data set for deposit-free rentals. Based on this multi-source behavioral data set, a relationship graph for deposit-free rentals is constructed, which includes business relationship edges and multiple types of nodes. Using the target user as an index, relationship mismatch identification is performed based on the feature differences between the two ends of the business relationship edges in the relationship graph, constructing a credit isomatch subgraph and a risk mismatch subgraph. Feature analysis is performed on the credit isomatch subgraph and the risk mismatch subgraph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics. Based on the user's stable credit characteristics and user risk anomaly characteristics, feature fusion is performed to obtain the user's comprehensive credit assessment result. Based on the user's comprehensive credit assessment result, deposit-free rental business management decision analysis is performed on the target user to obtain a deposit-free rental management strategy.

[0006] A second aspect of this application provides a credit-based deposit-free rental business management system, the system comprising:

[0007] The system comprises the following modules: a data acquisition module, a relationship graph construction module, and a decision analysis module. The data acquisition module iterates through the user set in the deposit-free rental business, collecting user information, rental order information, equipment information, address information, and payment account information for each user to construct a multi-source behavioral data set for deposit-free rentals. The relationship graph construction module constructs a relationship graph for the deposit-free rental business based on the multi-source behavioral data set, including business relationship edges and multiple types of nodes. The subgraph construction module uses the target user as an index and identifies relationship mismatches based on the feature differences between the two ends of the business relationship edges in the deposit-free rental business relationship graph, constructing a credit matching subgraph and a risk mismatch subgraph. The feature analysis module performs feature analysis on the credit matching subgraph and the risk mismatch subgraph respectively to obtain stable credit characteristics and abnormal risk characteristics of the user. The feature fusion module performs feature fusion based on the stable credit characteristics and abnormal risk characteristics of the user to obtain a comprehensive credit assessment result for the user. The decision analysis module performs deposit-free rental business management decision analysis on the target user based on the comprehensive credit assessment result to obtain a deposit-free rental management strategy.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] First, the user set in the deposit-free rental business is traversed, collecting user information, rental order information, equipment information, address information, and payment account information for each user to construct a multi-source behavioral data set for deposit-free rentals. Next, based on this multi-source behavioral data set, a relationship graph for the deposit-free rental business is constructed, including business relationship edges and various types of nodes. Further, using the target user as an index, relationship mismatch identification is performed based on the feature differences between the two ends of the business relationship edges in the deposit-free rental business relationship graph, constructing a credit isomatch subgraph and a risk mismatch subgraph. Subsequently, feature analysis is performed on the credit isomatch subgraph and the risk mismatch subgraph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics; feature fusion is then performed based on these characteristics to obtain a comprehensive user credit assessment result. Finally, based on the comprehensive user credit assessment result, deposit-free rental business management decision analysis is performed on the target user to obtain a deposit-free rental management strategy. This technology solves the technical problem of inaccurate user credit assessment in existing deposit-free leasing businesses, which leads to insufficient reliability of business management decisions. It achieves the technical effect of improving the accuracy of credit assessment and enhancing the reliability of business management decisions through multi-source behavioral data correlation analysis. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic diagram of a credit assessment-based deposit-free leasing business management method provided in this application embodiment;

[0012] Figure 2 This is a schematic diagram of a credit assessment-based deposit-free leasing business management system provided in an embodiment of this application.

[0013] Figure labeling: Data acquisition module 11, Relationship graph construction module 12, Subgraph construction module 13, Feature analysis module 14, Feature fusion module 15, Decision analysis module 16. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] Example 1, as Figure 1 As shown, this application provides a method for managing deposit-free leasing business based on credit assessment, wherein the method includes:

[0016] Traverse the user set in the deposit-free rental business, collect user information, rental order information, equipment information, address information and payment account information for each user, and construct a multi-source behavioral data set for deposit-free rental.

[0017] A user set is retrieved from the deposit-free rental business database, including current renters and historical renters. For each user in the user set, data is retrieved from the user master database, order management system database, equipment management system database, address management module, and payment settlement system database via data interfaces. The user information includes a unique user ID, registration time, real-name authentication status, historical credit rating, and account activity index. The rental order information includes order number, order time, rental period, performance status, whether it is overdue, overdue duration, number of defaults, number of cancellations, and early return flag. The equipment information includes a unique equipment identifier, equipment type, current equipment status, historical transfer records, and number of equipment anomaly flags. The address information includes a delivery address identifier, address usage frequency, number of users associated with the address, and historical abnormal address flags. The payment account information includes a unique payment account identifier, number of linked bank cards, number of users associated with the account, number of historical payment failures, and risk account identifier.

[0018] The collected raw data undergoes data cleaning and standardization, including removing missing value records, standardizing the time format, encoding discrete category data, and normalizing continuous numerical data. Multi-source data is then integrated based on the user's unique identifier to construct a multi-dimensional behavioral feature vector indexed by the user. This multi-dimensional behavioral feature vector for each user is stored in a structured data table, forming a multi-source behavioral data set for deposit-free rentals. This set uses the user as the primary key and includes user-dimensional features, order-dimensional features, device-dimensional features, address-dimensional features, and payment account-dimensional features. These features are used for subsequent construction of a deposit-free rental business relationship diagram and analysis of heterogeneous execution relationships.

[0019] Based on the aforementioned multi-source behavioral data set of deposit-free rentals, a deposit-free rental business relationship graph is constructed, wherein the deposit-free rental business relationship graph includes business relationship edges and multiple types of nodes.

[0020] Using a multi-source behavioral data set of deposit-free leasing as the input data source, entity extraction and node modeling are performed according to data field types. User nodes, device nodes, address nodes, and payment account nodes are constructed based on different entity types. Each node is assigned a unique identifier and corresponding node attribute features. The node attributes of user nodes include user credit feature vectors, historical performance indicators, and account activity indicators. The node attributes of device nodes include device type, historical leasing frequency, number of anomaly markers, and transfer path features. The node attributes of address nodes include address usage frequency, number of associated users, and anomaly address identifiers. The node attributes of payment account nodes include the number of accounts bound, historical payment success rate, and risk account markers.

[0021] Business relationship edges between nodes are established based on the associated fields in the multi-source behavioral data set; when a rental order record exists, a "rental relationship edge" is established between the corresponding user node and device node, and the rental time, performance status, and number of rentals are recorded as edge attributes; when a user fills in a delivery address in an order, a "use address relationship edge" is established between the user node and the address node, and the number of times the address is used and the most recent use time are recorded as edge attributes; when a user binds or uses a payment account to make a payment, a "payment association relationship edge" is established between the user node and the payment account node, and the payment frequency and the number of payment anomalies are recorded as edge attributes; when the same device, address, or payment account is used by multiple users, an indirect association path between users is formed based on shared entities.

[0022] The various types of nodes and business relationship edges are encapsulated into a unified graph structure to construct a deposit-free rental business relationship graph containing a node set N and an edge set E. The node set N includes a user node set, a device node set, an address node set, and a payment account node set, and the edge set E includes rental relationship edges, address association edges, and payment association edges. The business relationship graph can be stored using an adjacency list or adjacency matrix structure for subsequent K-order association extraction and relationship mismatch identification and analysis.

[0023] Furthermore, the various types of nodes in the deposit-free rental business relationship diagram include user nodes, device nodes, address nodes, and payment account nodes; the business relationship edges in the deposit-free rental business relationship diagram are used to describe the business relationships between user nodes, device nodes, address nodes, and payment account nodes.

[0024] Based on the entity fields in the multi-source behavioral data set of deposit-free leasing, node types are classified and instantiated. Among them, user nodes are used to represent natural persons or legal entities participating in deposit-free leasing business. Each user node uses the user's unique identifier ID as the node primary key and is associated with its corresponding credit feature vector. Device nodes are used to represent leasing equipment participating in the leasing process. Each device node uses the device's unique identifier as the node primary key and is associated with attributes such as device type, historical transfer count, and abnormal records. Address nodes are used to represent order delivery addresses or equipment delivery addresses. Each address node uses the address's unique identifier as the node primary key and is associated with the address usage frequency and the number of associated users. Payment account nodes are used to represent the payment accounts bound or used by users. Each payment account node uses the account's unique identifier as the node primary key and is associated with payment frequency and risk marker attributes.

[0025] The business relationship edges are constructed based on business interaction records. When a user generates a rental order, a rental relationship edge is established between the user node and the device node, and the order number, rental period, performance status, and number of performances are used as edge attributes. When a user uses a specific address in an order, an address association edge is established between the user node and the address node, and the number of times the address is used and the most recent usage time are recorded as edge attributes. When a user completes a payment through a payment account, a payment association edge is established between the user node and the payment account node, and the payment success rate and the number of payment anomalies are recorded as edge attributes. When the same device node, address node, or payment account node is connected by multiple user nodes, an indirect cross-user association path is formed, which is used to characterize the potential association risk propagation structure.

[0026] Using the target user as an index, the relationship heteromatch is identified based on the feature differences between the two ends of the business relationship edge in the deposit-free rental business relationship graph, and a credit isomatch subgraph and a risk heteromatch subgraph are constructed.

[0027] Furthermore, using the target user as an index, and based on the feature differences between the two ends of the business relationship edge in the deposit-free rental business relationship graph, relationship mismatch identification is performed, and a credit isomatch subgraph and a risk mismatch subgraph are constructed, including:

[0028] Using the target user as an index, K-order association extraction is performed on the deposit-free rental business relationship graph to obtain a related deposit-free rental business relationship graph, where K is a positive integer greater than or equal to 2; according to preset credit indicators, the nodes in the related deposit-free rental business relationship graph are traversed to extract node credit features to obtain a node credit feature set; taking the target node corresponding to the target user in the related deposit-free rental business relationship graph as the starting point, and combining the node credit feature set, relationship heteromatch identification is performed to construct the credit isomatch subgraph and the risk heteromatch subgraph.

[0029] Preferably, starting with the user node corresponding to the target user, K-order association extraction is performed in the deposit-free rental business relationship graph to obtain the associated deposit-free rental business relationship graph. First-order associations are used to extract adjacent nodes and their business relationship edges that have a direct business relationship with the target node; second-order associations are used to extract nodes indirectly connected through first-order nodes and their business relationship edges; third-order and higher-order associations are used to extract indirect associated nodes at more distant levels. K is a positive integer greater than or equal to 2, used to control the association propagation depth, and the value of K is set according to the business risk identification requirements and computational complexity requirements. All associated node sets and business relationship edge sets within the K-order range are obtained through breadth-first traversal or depth-first traversal, forming the target user association subgraph. According to preset credit indicators, node credit features are extracted from various nodes in the associated deposit-free rental business relationship graph. These preset credit indicators include performance rate, overdue rate, number of defaults, order cancellation frequency, and abnormal behavior markers. For each node, indicators are calculated and normalized based on its historical behavior data to generate a corresponding node credit feature vector. All node credit feature vectors are then aggregated to construct a node credit feature set. Starting from the target node corresponding to the target user, the business relationship edges in the associated deposit-free rental business relationship graph are traversed. For each business relationship edge, the node credit feature vectors of its two endpoints are extracted and their dimensions are aligned and their differences are calculated to obtain the feature difference of the business relationship edge. The feature difference is compared with preset matching thresholds and dismatch thresholds. When the feature difference is less than or equal to the matching threshold, the corresponding business relationship edge and its associated nodes are assigned to a credit matching subgraph. When the feature difference is greater than the dismatch threshold, the corresponding business relationship edge and its associated nodes are assigned to a risk dismatch subgraph, thus completing the identification of relationship dismatch and subgraph construction.

[0030] Furthermore, the preset credit indicators include the normal fulfillment rate of historical leasing orders, the proportion of overdue orders, the number of defaulted orders, the proportion of early returned orders, and the frequency of order cancellations per unit time.

[0031] Based on the historical rental order records in the aforementioned deposit-free rental multi-source behavioral data set, statistical analysis is performed on the order data of each user within a preset statistical period T; wherein, the statistical period T is a preset time window, such as the most recent 6 months, 12 months, or cumulative historical period.

[0032] The normal performance rate is calculated as follows:

[0033] Normal fulfillment rate = Number of orders completed normally / Total number of orders within the statistical period; where, orders completed normally are those that have not been overdue or defaulted and have been returned on time as agreed.

[0034] The percentage of overdue orders is calculated as follows:

[0035] Overdue order percentage = number of overdue orders / total number of orders in the statistical period; the number of defaulted orders is the total number of orders marked as defaulted by the system in the statistical period.

[0036] The proportion of orders to be returned early is calculated as follows:

[0037] Early return rate = Number of early returned orders / Total number of orders within the statistical period.

[0038] The order cancellation frequency per unit time is calculated as follows:

[0039] Order cancellation frequency = number of canceled orders within the statistical period / length of the statistical period; where the length of the statistical period can be in days, weeks or months.

[0040] Furthermore, taking the target node corresponding to the target user in the associated deposit-free rental business relationship graph as the starting point, and combining the node credit feature set to perform relationship heteromatch identification, the credit isomatch subgraph and the risk heteromatch subgraph are constructed, including:

[0041] Traverse the business relationship edges in the associated deposit-free rental business relationship graph, and perform feature difference analysis on the nodes at both ends of the business relationship edge in combination with the node credit feature set to obtain the corresponding business relationship edge feature difference results; based on the business relationship edge feature difference results, perform relationship mismatch identification on the business relationship edge, and map the business relationship edge to the credit isomatch subgraph and the risk mismatch subgraph respectively according to the relationship mismatch identification results.

[0042] Traverse the business relationship edges in the associated deposit-free rental business relationship graph; for each business relationship edge, extract the node credit feature vectors of the two ends of the business relationship edge in the node credit feature set; perform dimension alignment processing on the node credit feature vectors of the two ends of the business relationship edge, and perform multi-dimensional feature difference calculation to obtain the corresponding business relationship edge feature difference value; the difference calculation method includes weighted summation of the absolute differences of each dimension of credit features or generating a comprehensive difference using Euclidean distance calculation method.

[0043] Relationship mismatch identification is performed based on the feature difference value of the business relationship edge; the feature difference value is compared with a preset mismatch determination threshold. When the feature difference value is greater than or equal to the mismatch determination threshold, the business relationship edge is determined to be a risk mismatch edge and added to the risk mismatch subgraph; when the feature difference value is less than the mismatch determination threshold, the business relationship edge is determined to be a credit matching edge and added to the credit matching subgraph.

[0044] After completing the identification of heteromatches of all business relationship edges, the associated nodes are synchronously mapped according to the attribution results of each business relationship edge to form the credit isomatch subgraph and the risk heteromatch subgraph, which are used for subsequent stable credit feature analysis and abnormal risk feature extraction.

[0045] Furthermore, by traversing the business relationship edges in the aforementioned associated deposit-free rental business relationship graph and combining the node credit feature set, feature difference analysis is performed on the nodes at both ends of the business relationship edge to obtain the corresponding business relationship edge feature difference results, including:

[0046] Extract the node credit features of the two ends of each business relationship edge from the node credit feature set to construct a matching node credit feature set; perform dimensional alignment and difference calculation on the matching node credit feature set to obtain a multidimensional matching node credit feature set difference set; use the multidimensional matching node credit feature set difference set as the feature difference result of the business relationship edge.

[0047] Preferably, the node credit feature vectors corresponding to the two ends of each business relationship edge are extracted from the node credit feature set to construct a matching node credit feature group set. Each matching node credit feature group includes a first node credit feature vector and a second node credit feature vector, where each node credit feature vector consists of multiple credit indicator dimensions, each including at least one quantifiable credit indicator. Dimensional alignment is performed on each set of node credit feature vectors in the matching node credit feature group set to ensure a one-to-one correspondence between each credit indicator dimension. A difference value is calculated for each credit indicator dimension, where the difference value is the result of the difference function of the corresponding indicators at both ends of the node. The difference values ​​of multiple indicators under the same credit feature dimension are weighted and calculated to obtain the intra-group difference value for that credit feature dimension. Then, the intra-group difference values ​​of all credit feature dimensions are weighted and fused according to a preset weight coefficient to generate a comprehensive difference value for the corresponding business relationship edge, thus forming a multi-dimensional matching node credit feature group difference value set. This multi-dimensional matching node credit feature group difference value set is used as the feature difference result of the business relationship edge for subsequent relationship mismatch identification and subgraph construction.

[0048] Feature analysis is performed on the credit isogamete graph and the risk heterogamete graph respectively to obtain the user's stable credit characteristics and user risk abnormal characteristics.

[0049] Furthermore, feature analysis is performed on the credit isogamete graph and the risk heterogamete graph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics, including:

[0050] By traversing the credit isomatched subgraph, we can analyze performance behavior, identity consistency, and relationship stability to obtain stable credit characteristics that represent the long-term credit performance of the target user. By traversing the risk heteromatched subgraph, we can analyze behavioral differences and relationship conflicts to obtain abnormal risk characteristics that represent the abnormal behavior patterns of the target user.

[0051] The credit isomatch subgraph is traversed to perform performance behavior, identity consistency, and relationship stability analysis to obtain stable credit characteristics that characterize the long-term credit performance of the target user. The performance behavior analysis includes statistically analyzing the target user's normal performance rate, average performance period, number of defaults, and early repayment rate within a preset statistical period in the credit isomatch subgraph, and normalizing these indicators. The identity consistency analysis includes statistically analyzing the consistency of device type, address usage, and payment account stability used by the target user in different rental periods, obtaining an identity consistency index by calculating the consistency ratio or reuse rate. The relationship stability analysis includes statistically analyzing the number of consecutive associations, association duration, and repetitive association frequency between the target user and associated nodes in the credit isomatch subgraph, comparing these with a preset stability threshold to generate a relationship stability index. The performance behavior index, identity consistency index, and relationship stability index are then weighted and fused to form the user's stable credit characteristics.

[0052] The risk heteromatch subgraph is traversed to perform behavioral difference and relationship conflict analysis, thereby obtaining user risk anomaly features that characterize the abnormal behavior patterns of the target user. The behavioral difference analysis includes statistically analyzing the difference in fulfillment rate, the difference in default frequency, and the difference in order cancellation frequency between the target user and high-difference nodes, and calculating the mean or extreme value of the difference. The relationship conflict analysis includes statistically analyzing the number of device sharing conflicts, the number of address reuse anomalies, and the number of cross-uses of payment accounts in the risk heteromatch subgraph, and calculating the conflict intensity coefficient. The behavioral difference indicators and relationship conflict indicators are then normalized and weighted and fused to generate user risk anomaly features.

[0053] Furthermore, by traversing the credit isogamete subgraph to analyze performance behavior, identity consistency, and relationship stability, stable credit characteristics representing the long-term credit performance of the target user are obtained, including:

[0054] Traverse the credit isomatch subgraph to extract performance behavior data over multiple lease periods, perform stability analysis, and obtain performance behavior characteristics; extract identity information data, device usage data, and address usage data from the credit isomatch subgraph, perform joint consistency analysis, and obtain identity consistency characteristics; count the number of continuous associations in the credit isomatch subgraph, and compare the statistical results with a preset number threshold to obtain relationship stability characteristics; summarize the performance behavior characteristics, identity consistency characteristics, and relationship stability characteristics to obtain the user's stable credit characteristics.

[0055] The credit isomatch graph is traversed to extract the performance behavior data of target users over multiple rental periods. The performance behavior data includes the number of normally performed orders, the number of overdue orders, the number of defaulted orders, and the number of early returned orders. Time series stability analysis is performed on the above performance behavior data in a preset statistical period to calculate the mean performance rate, the standard deviation of the performance rate, and the coefficient of variation of the performance rate. The stability of performance behavior is determined based on the coefficient of variation of the performance rate. The smaller the coefficient of variation, the lower the fluctuation of performance behavior and the higher the stability, thereby obtaining the performance behavior characteristics.

[0056] Extract identity information data, device usage data, and address usage data from the credit isomatch subgraph. The identity information data includes the consistency status of real-name authentication and the number of historical account changes. The device usage data includes the device type reuse rate and device replacement frequency. The address usage data includes the address reuse rate and address change frequency. Perform joint consistency analysis on the above multidimensional data, calculate the consistency ratio of each dimension, and generate a comprehensive consistency coefficient through weighted fusion. The higher the consistency coefficient, the more stable the identity usage behavior, thereby obtaining identity consistency characteristics.

[0057] The number of relation edges forming continuous association paths with the target user in the credit isomatch subgraph is statistically analyzed. A continuous association path is defined as a sequence of relation edges that occur adjacently in the time dimension and whose associated nodes remain consistent. When the number of continuous association edges is greater than or equal to a preset continuous threshold, it is determined to be a stable association structure. The ratio of the number of continuous association edges to the total number of association edges is calculated to obtain the relationship stability coefficient. The larger the relationship stability coefficient, the higher the proportion of continuous associations and the more stable the association structure, thereby obtaining the relationship stability characteristics.

[0058] The performance behavior characteristics, identity consistency characteristics, and relationship stability characteristics are normalized and then weighted and fused according to preset weight coefficients to generate user stable credit characteristics, which are used to characterize the long-term credit stability of the target user.

[0059] Furthermore, by traversing the aforementioned risk heteromatch subgraph to perform behavioral difference and relationship conflict analysis, user risk anomaly characteristics representing the abnormal behavior patterns of target users are obtained, including:

[0060] The credit isogamete graph is traversed to count the frequency of abnormal performance behavior, and the frequency of abnormal performance behavior is obtained. The credit isogamete graph is traversed to analyze the degree of device sharing conflict and the degree of address reuse anomaly, and the strength of relationship conflict is obtained. The frequency of abnormal performance behavior and the strength of relationship conflict are summarized to obtain the abnormal characteristics of user risk.

[0061] Traverse the risk heteromatch subgraph to extract performance behavior data between the target user and high-difference nodes, and count the number of abnormal performances within a preset statistical period; the abnormal performance behaviors include overdue orders, defaulted orders, and abnormally canceled orders; calculate the abnormal performance frequency = number of abnormal orders / length of statistical period to obtain the abnormal performance frequency index; at the same time, calculate the difference between the abnormal performance rate and the historical average performance rate of the target user to characterize the degree of fluctuation of abnormal behavior.

[0062] The shared entity nodes in the risk heterogeneous subgraph are traversed, and device sharing conflict analysis and address reuse anomaly analysis are performed. The degree of device sharing conflict is calculated by counting the number of times the same device node is associated with multiple high-risk users in the risk heterogeneous subgraph, and the device conflict coefficient is calculated as: number of conflict associations / total number of device leases. The degree of address reuse anomaly is calculated by counting the number of times the same address node is reused by multiple abnormal users in the risk heterogeneous subgraph, and the address anomaly coefficient is calculated as: number of abnormal associated users / total number of address associated users. The device conflict coefficient and the address anomaly coefficient are weighted and fused to generate a relationship conflict intensity index.

[0063] The abnormal frequency index of performance behavior and the intensity index of relationship conflict are normalized and then weighted and fused according to preset weights to generate abnormal user risk features. The abnormal user risk features are used to characterize the abnormal behavior patterns and potential associated risk levels of target users in the risk heterogeneous structure.

[0064] Based on the user's stable credit characteristics and abnormal risk characteristics, feature fusion is performed to obtain the user's comprehensive credit assessment result.

[0065] The stable credit characteristics and abnormal risk characteristics of users are normalized to ensure that the feature values ​​of each dimension are within a uniform range. The stable credit characteristics of users are weighted and summed according to preset weight coefficients to obtain a stable credit score. At the same time, the abnormal risk characteristics of users are weighted and summed according to preset weight coefficients to obtain a risk abnormality score.

[0066] The stable credit score and the risk anomaly score are integrated for calculation, with the stable credit score serving as a positive contributor and the risk anomaly score as a negative corrective factor. A weighted combination of the two is used to generate a comprehensive user credit assessment result. The sum of the stable credit feature weight and the risk anomaly feature weight is 1, with the stable credit feature weight being higher than the risk anomaly feature weight to reflect the dominant role of long-term stable performance. The weighting coefficients are determined based on historical sample data statistics or risk loss regression analysis results; when historical risk loss levels increase, the risk anomaly feature weight is appropriately increased; when a user's performance stability is outstanding, the stable credit feature weight is increased.

[0067] Based on the comprehensive credit assessment results of the users, the management decision analysis of deposit-free rental business is carried out on the target users to obtain deposit-free rental management strategies.

[0068] A credit rating threshold classification table is constructed based on the user's comprehensive credit assessment results. The threshold classification table is divided into three levels according to the comprehensive credit assessment score range, including the deposit-free approval range, the conditional deposit-free range, and the high-risk restriction range. The threshold for each range is set based on the historical default rate statistics and the level of risk loss.

[0069] When the user's comprehensive credit assessment result is within the deposit-free approval range, the target user is deemed to have stable creditworthiness, a deposit-free leasing approval strategy is generated, and the user is allowed to directly apply for deposit-free leasing business; when the user's comprehensive credit assessment result is within the conditional deposit-free range, the target user is deemed to have certain risk fluctuations, and a risk monitoring strategy is generated, including but not limited to shortening the leasing period, restricting the types of equipment that can be leased, setting dynamic risk monitoring markers, or raising the order review level; when the user's comprehensive credit assessment result is within the high-risk restriction range, the target user is deemed to have a high default risk, and a risk control strategy is generated, including requiring the payment of a security deposit, restricting deposit-free privileges, or suspending leasing eligibility.

[0070] After the deposit-free rental management strategy is generated, the corresponding strategy is written into the business management system and the user risk status label is updated synchronously for subsequent order approval processes and risk monitoring module calls, thereby realizing dynamic business management decision-making closed-loop control based on credit assessment results.

[0071] In summary, the embodiments of this application have at least the following technical effects:

[0072] First, the user set in the deposit-free rental business is traversed, collecting user information, rental order information, equipment information, address information, and payment account information for each user to construct a multi-source behavioral data set for deposit-free rentals. Next, based on this multi-source behavioral data set, a relationship graph for the deposit-free rental business is constructed, including business relationship edges and various types of nodes. Further, using the target user as an index, relationship mismatch identification is performed based on the feature differences between the two ends of the business relationship edges in the deposit-free rental business relationship graph, constructing a credit isomatch subgraph and a risk mismatch subgraph. Subsequently, feature analysis is performed on the credit isomatch subgraph and the risk mismatch subgraph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics; feature fusion is then performed based on these characteristics to obtain a comprehensive user credit assessment result. Finally, based on the comprehensive user credit assessment result, deposit-free rental business management decision analysis is performed on the target user to obtain a deposit-free rental management strategy. This technology solves the technical problem of inaccurate user credit assessment in existing deposit-free leasing businesses, which leads to insufficient reliability of business management decisions. It achieves the technical effect of improving the accuracy of credit assessment and enhancing the reliability of business management decisions through multi-source behavioral data correlation analysis.

[0073] Example 2 is based on the same inventive concept as the credit assessment-based deposit-free leasing business management method in the previous examples, such as... Figure 2 As shown, this application provides a credit-based deposit-free rental business management system, wherein the system includes:

[0074] Data Acquisition Module 11: Traverses the user set in the deposit-free rental business, collects user information, rental order information, equipment information, address information, and payment account information corresponding to each user, and constructs a multi-source behavioral data set for deposit-free rental; Relationship Graph Construction Module 12: Based on the multi-source behavioral data set for deposit-free rental, constructs a relationship graph for the deposit-free rental business, wherein the relationship graph includes business relationship edges and multiple types of nodes; Subgraph Construction Module 13: Using the target user as an index, identifies the mismatch of relationships based on the feature differences between the two ends of the business relationship edges in the relationship graph, and constructs a credit isomatch subgraph and a risk mismatch subgraph; Feature Analysis Module 14: Performs feature analysis on the credit isomatch subgraph and the risk mismatch subgraph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics; Feature Fusion Module 15: Performs feature fusion based on the user's stable credit characteristics and user risk anomaly characteristics to obtain the user's comprehensive credit assessment result; Decision Analysis Module 16: Based on the user's comprehensive credit assessment result, performs deposit-free rental business management decision analysis on the target user to obtain a deposit-free rental management strategy.

[0075] Furthermore, the relationship graph construction module 12 is used to perform the following methods:

[0076] The deposit-free rental business relationship diagram includes multiple types of nodes, such as user nodes, device nodes, address nodes, and payment account nodes; the business relationship edges in the deposit-free rental business relationship diagram are used to describe the business relationships between user nodes, device nodes, address nodes, and payment account nodes.

[0077] Furthermore, the subgraph construction module 13 is used to perform the following method:

[0078] Using the target user as an index, K-order association extraction is performed on the deposit-free rental business relationship graph to obtain a related deposit-free rental business relationship graph, where K is a positive integer greater than or equal to 2; according to preset credit indicators, the nodes in the related deposit-free rental business relationship graph are traversed to extract node credit features to obtain a node credit feature set; taking the target node corresponding to the target user in the related deposit-free rental business relationship graph as the starting point, and combining the node credit feature set, relationship heteromatch identification is performed to construct the credit isomatch subgraph and the risk heteromatch subgraph.

[0079] Furthermore, the subgraph construction module 13 is used to perform the following method:

[0080] The preset credit indicators include the normal fulfillment rate of historical lease orders, the proportion of overdue orders, the number of defaulted orders, the proportion of early returned orders, and the frequency of order cancellations per unit time.

[0081] Furthermore, the subgraph construction module 13 is used to perform the following method:

[0082] Traverse the business relationship edges in the associated deposit-free rental business relationship graph, and perform feature difference analysis on the nodes at both ends of the business relationship edge in combination with the node credit feature set to obtain the corresponding business relationship edge feature difference results; based on the business relationship edge feature difference results, perform relationship mismatch identification on the business relationship edge, and map the business relationship edge to the credit isomatch subgraph and the risk mismatch subgraph respectively according to the relationship mismatch identification results.

[0083] Furthermore, the subgraph construction module 13 is used to perform the following method:

[0084] Extract the node credit features of the two ends of each business relationship edge from the node credit feature set to construct a matching node credit feature set; perform dimensional alignment and difference calculation on the matching node credit feature set to obtain a multidimensional matching node credit feature set difference set; use the multidimensional matching node credit feature set difference set as the feature difference result of the business relationship edge.

[0085] Furthermore, the feature analysis module 14 is used to perform the following methods:

[0086] By traversing the credit isomatched subgraph, we can analyze performance behavior, identity consistency, and relationship stability to obtain stable credit characteristics that represent the long-term credit performance of the target user. By traversing the risk heteromatched subgraph, we can analyze behavioral differences and relationship conflicts to obtain abnormal risk characteristics that represent the abnormal behavior patterns of the target user.

[0087] Furthermore, the feature analysis module 14 is used to perform the following methods:

[0088] Traverse the credit isomatch subgraph to extract performance behavior data over multiple lease periods, perform stability analysis, and obtain performance behavior characteristics; extract identity information data, device usage data, and address usage data from the credit isomatch subgraph, perform joint consistency analysis, and obtain identity consistency characteristics; count the number of continuous associations in the credit isomatch subgraph, and compare the statistical results with a preset number threshold to obtain relationship stability characteristics; summarize the performance behavior characteristics, identity consistency characteristics, and relationship stability characteristics to obtain the user's stable credit characteristics.

[0089] Furthermore, the feature analysis module 14 is used to perform the following methods:

[0090] The credit isogamete graph is traversed to count the frequency of abnormal performance behavior, and the frequency of abnormal performance behavior is obtained. The credit isogamete graph is traversed to analyze the degree of device sharing conflict and the degree of address reuse anomaly, and the strength of relationship conflict is obtained. The frequency of abnormal performance behavior and the strength of relationship conflict are summarized to obtain the abnormal characteristics of user risk.

[0091] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for managing deposit-free leasing business based on credit assessment, characterized in that, The method includes: Traverse the user set in the deposit-free rental business, collect user information, rental order information, equipment information, address information and payment account information for each user, and construct a multi-source behavioral data set for deposit-free rental; Based on the aforementioned multi-source behavioral data set of deposit-free rental, a deposit-free rental business relationship graph is constructed, wherein the deposit-free rental business relationship graph includes business relationship edges and multiple types of nodes; Using the target user as an index, the relationship heteromatch is identified based on the feature differences between the two ends of the business relationship edge in the deposit-free rental business relationship graph, and a credit isomatch subgraph and a risk heteromatch subgraph are constructed. Feature analysis is performed on the credit isogamete graph and the risk heterogamete graph respectively to obtain the user's stable credit characteristics and user risk abnormal characteristics; Based on the user's stable credit characteristics and user risk anomaly characteristics, feature fusion is performed to obtain the user's comprehensive credit assessment result; Based on the comprehensive credit assessment results of the users, the management decision analysis of deposit-free rental business is carried out on the target users to obtain deposit-free rental management strategies.

2. The method for managing deposit-free leasing business based on credit assessment as described in claim 1, characterized in that, The relationship diagram of the deposit-free rental business includes multiple types of nodes, such as user nodes, device nodes, address nodes, and payment account nodes. The business relationship edges in the deposit-free rental business relationship diagram are used to describe the business relationships between user nodes, device nodes, address nodes, and payment account nodes.

3. The method for managing deposit-free leasing business based on credit assessment as described in claim 1, characterized in that, Using the target user as an index, and based on the feature differences between the two ends of the business relationship edge in the deposit-free rental business relationship graph, relationship mismatch identification is performed, and a credit isomatch subgraph and a risk mismatch subgraph are constructed, including: Using the target user as the index, K-order association extraction is performed on the deposit-free rental business relationship graph to obtain the associated deposit-free rental business relationship graph, where K is a positive integer greater than or equal to 2; Based on preset credit indicators, the nodes in the associated deposit-free rental business relationship graph are traversed to extract node credit features and obtain a set of node credit features. Starting from the target node corresponding to the target user in the associated deposit-free rental business relationship graph, and combining the node credit feature set to perform relationship heteromatch identification, the credit isomatch subgraph and the risk heteromatch subgraph are constructed.

4. The method for managing deposit-free leasing business based on credit assessment as described in claim 3, characterized in that, The preset credit indicators include the normal fulfillment rate of historical lease orders, the proportion of overdue orders, the number of defaulted orders, the proportion of early returned orders, and the frequency of order cancellations per unit time.

5. The method for managing deposit-free leasing business based on credit assessment as described in claim 3, characterized in that, Starting from the target node corresponding to the target user in the associated deposit-free rental business relationship graph, and combining the node credit feature set to perform relationship heteromatch identification, the credit isomatch subgraph and the risk heteromatch subgraph are constructed, including: Traverse the business relationship edges in the associated deposit-free rental business relationship graph, and combine the node credit feature set to perform feature difference analysis on the nodes at both ends of the business relationship edge to obtain the corresponding business relationship edge feature difference results. Based on the difference results of the business relationship edge features, the business relationship edge is identified for heteromatch, and according to the heteromatch identification results, the business relationship edge is mapped to the credit isomatch subgraph and the risk heteromatch subgraph respectively.

6. The method for managing deposit-free leasing business based on credit assessment as described in claim 5, characterized in that, Traverse the business relationship edges in the associated deposit-free rental business relationship graph, and combine the node credit feature set to perform feature difference analysis on the nodes at both ends of the business relationship edge to obtain the corresponding business relationship edge feature difference results, including: Extract the node credit features of the two ends of each business relationship edge from the node credit feature set, and construct a matching node credit feature set; Dimension alignment and difference calculation are performed on the set of credit feature groups of matching nodes respectively to obtain a multidimensional set of differences in credit feature groups of matching nodes. The set of differences in the credit feature groups of the multidimensional matching nodes is used as the result of the difference in the features of the business relationship edges.

7. The method for managing deposit-free leasing business based on credit assessment as described in claim 1, characterized in that, Feature analysis is performed on the credit isogamete graph and the risk heterogamete graph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics, including: By traversing the credit isogamete graph, we can analyze performance behavior, identity consistency, and relationship stability to obtain stable credit characteristics that represent the long-term credit performance of the target user. By traversing the risk heterogametic subgraph, behavioral differences and relationship conflicts are analyzed to obtain user risk anomaly characteristics that represent the abnormal behavior patterns of target users.

8. The method for managing deposit-free leasing business based on credit assessment as described in claim 7, characterized in that, By traversing the credit isogamete graph to analyze performance behavior, identity consistency, and relationship stability, stable credit characteristics representing the long-term credit performance of the target user are obtained, including: Traverse the credit isogamete graph, extract performance behavior data over multiple lease periods, perform stability analysis, and obtain performance behavior characteristics; Extract identity information data, device usage data, and address usage data from the credit isomatch subgraph, perform joint consistency analysis, and obtain identity consistency features; The number of continuous associations in the credit isogamete graph is counted, and the statistical results are compared with a preset threshold to obtain the relationship stability characteristics. By summarizing the performance behavior characteristics, identity consistency characteristics, and relationship stability characteristics, the user's stable credit characteristics are obtained.

9. The method for managing deposit-free leasing business based on credit assessment as described in claim 7, characterized in that, By traversing the aforementioned risk heterogametic subgraph to perform behavioral difference and relationship conflict analysis, user risk anomaly characteristics representing the abnormal behavior patterns of target users are obtained, including: Traverse the credit isogamete graph to count the frequency of abnormal performance behavior and obtain the frequency of abnormal performance behavior; By traversing the credit isogamete graph, the degree of device sharing conflict and address reuse anomaly is analyzed to obtain the strength of relationship conflict. By summarizing the frequency of abnormal performance behaviors and the intensity of relationship conflicts, the abnormal risk characteristics of the users are obtained.

10. A credit-based deposit-free leasing business management system, characterized in that, The system is used to implement the credit assessment-based deposit-free leasing business management method according to any one of claims 1-9, the system comprising: Data acquisition module: Traverses the user set in the deposit-free rental business, collects user information, rental order information, equipment information, address information and payment account information for each user, and constructs a multi-source behavioral data set for deposit-free rental; Relationship Graph Construction Module: Based on the multi-source behavioral data set of deposit-free rental, construct a relationship graph for deposit-free rental business, wherein the relationship graph for deposit-free rental business includes business relationship edges and multiple types of nodes; Subgraph construction module: Using the target user as the index, it identifies the relationship mismatch based on the feature differences between the two ends of the business relationship edge in the deposit-free rental business relationship graph, and constructs a credit isomatch subgraph and a risk mismatch subgraph; Feature analysis module: Performs feature analysis on the credit isogamete graph and risk heterogamete graph respectively to obtain the user's stable credit characteristics and user risk anomaly characteristics; Feature fusion module: Performs feature fusion based on the user's stable credit characteristics and user risk anomaly characteristics to obtain the user's comprehensive credit assessment result; Decision Analysis Module: Based on the comprehensive credit assessment results of the users, the module performs decision analysis on the management of deposit-free rental services for target users and obtains deposit-free rental management strategies.

Images