Domestic credit risk conduction analysis method and system based on knowledge graph

Abstract

The invention relates to the technical field of financial risks, and particularly discloses a knowledge graph-based family credit risk conduction analysis method and system, and the method comprises the steps: obtaining the family member data of a target family, and calculating the personal risk value of each family member of the target family; acquiring family communication data of a target family, constructing a member knowledge graph of the target family, and constructing a risk conduction matrix of the target family; determining risk conduction sub-data of each family member of the target family; and determining risk conduction data of the target family based on the member knowledge graph of the target family, the personal risk value of each family member, the risk conduction sub-data and the risk conduction matrix. According to the method, networked and systematic analysis of family credit risks can be realized based on the knowledge graph, hidden relatives associated risks are accurately mined, the comprehensiveness and accuracy of risk pre-judgment are improved, high-value decision support is provided for family credit risk management and control, and risk analysis requirements of a complex family structure are met.

Description

A knowledge graph-based method and system for analyzing the transmission of household credit risk Technical Field

[0001] This invention relates to the field of financial risk technology, and in particular to a method and system for analyzing the transmission of household credit risk based on knowledge graphs. Background Technology

[0002] Traditional family credit risk analysis has long relied on credit data from single entities, focusing on individual credit assessments while neglecting the risk transmission effects of kinship and economic transactions among family members. This leads to one-sided and delayed risk predictions. Early financial credit assessments were mostly targeted at businesses or individual individuals, lacking a systemic risk analysis framework at the family level. With the increase in joint family lending and borrowing, chain reactions caused by risk transmission among relatives have become frequent. In recent years, knowledge graph technology has been gradually applied in the field of financial risk analysis, but in the family credit scenario, a risk transmission analysis solution integrating generational hierarchy and multi-dimensional transaction data has not yet been developed, making it difficult to adapt to the risk analysis needs of complex family structures.

[0003] Therefore, this invention proposes a method and system for analyzing the transmission of family credit risk based on knowledge graphs. Summary of the Invention

[0004] This invention provides a knowledge graph-based method and system for analyzing family credit risk transmission. By acquiring data on family members and their interactions within a target family, the system calculates the individual risk value for each family member, constructs a knowledge graph of the target family members, and builds a risk transmission matrix. This determines the risk transmission sub-data for each family member and the overall risk transmission data for the target family. Leveraging the knowledge graph, the system enables networked and systematic analysis of family credit risk, accurately uncovering hidden kinship-related risks, improving the comprehensiveness and accuracy of risk prediction, providing high-value decision support for family credit risk management, and adapting to the risk analysis needs of complex family structures.

[0005] This invention provides a knowledge graph-based method for analyzing the transmission of family credit risk, comprising: S1: acquiring family member data of a target family, and calculating the individual risk value of each family member based on the family member data; S2: acquiring family transaction data of the target family, constructing a family member knowledge graph of the target family based on the family member data and the family transaction data, and constructing a risk transmission matrix of the target family; S3: determining the risk transmission sub-data of each family member based on the family member knowledge graph and the risk transmission matrix; S4: determining the risk transmission data of the target family based on the family member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix.

[0006] Preferably, a knowledge graph-based method for analyzing the transmission of family credit risk involves acquiring data on family members of a target family, including: acquiring the names, generational rankings, and personal credit data of each family member. The personal credit data includes personal credit score, personal assets, periodic income and liabilities over multiple specified time periods, and historical overdue data. The historical overdue data includes sub-data on multiple past overdue periods, which includes overdue type tags, overdue start date, overdue end date, overdue amount, and overdue status tags. Overdue type tags include credit cards, consumer loans, mortgages, business loans, etc., and overdue status tags include settled, written off, and repaid. Based on the names, generational rankings, and personal credit data of all family members in the target family, the family member data of the target family is determined.

[0007] Preferably, a knowledge graph-based family credit risk transmission analysis method calculates the individual risk value of each family member of the target family based on the family member data of the target family, including: calculating the credit risk value, cycle risk value, asset risk value, and delinquency risk value of each family member of the target family based on the individual credit data of each family member in the family member data of the target family, and calculating the individual risk value of each family member of the target family.

[0008] Preferably, a knowledge graph-based method for analyzing the transmission of family credit risk involves acquiring family transaction data of a target family, including: acquiring the kinship relationship and transaction data between every two family members of the target family; the kinship transaction data includes multiple sub-data of kinship transactions for all specified time periods or is empty; the sub-data of kinship transactions includes transaction type tags, transaction amounts, transaction start dates, and transaction end dates; the transaction type tags include joint lending, joint mortgage, guarantee, and daily turnover; the transaction type tags are the names of the guarantor and the guaranteed party at the time of guarantee, and the transaction type tags are the names of the transferor and the recipient at the time of daily turnover; and based on the kinship transaction data between all family members of the target family, the family transaction data of the target family is determined.

[0009] Preferably, a knowledge graph-based method for analyzing family credit risk transmission involves constructing a family member knowledge graph and a risk transmission matrix for the target family based on family member data and family transaction data. This includes: constructing the family member knowledge graph based on the member names, generational ranks, and kinship relationships between any two family members in the family transaction data; calculating the transmission amount of each family member to each other family member's kinship transaction data if the kinship transaction data is not empty; calculating the risk transmission value of each family member to each other family member's kinship transaction data if the kinship transaction data is empty; determining the risk transmission value between any two family members of the target family to be 0; and constructing the risk transmission matrix for the target family based on the risk transmission values ​​among all family members.

[0010] Preferably, a knowledge graph-based method for analyzing family credit risk transmission involves determining risk transmission sub-data for each family member based on the target family's member knowledge graph and risk transmission matrix. This includes: based on the target family's member knowledge graph, extending two generational levels vertically and horizontally along the same generational level of the member knowledge graph to determine multiple risk transmission paths for each family member; if any two family members in a risk transmission path have a risk transmission value of 0, the path label for that family member's risk transmission path is determined to be discarded; if no two family members in a risk transmission path have a risk transmission value of 0, the path label for that family member's risk transmission path is determined to be normal; and based on all risk transmission paths with normal path labels for each family member, the risk transmission sub-data for each family member is determined.

[0011] Preferably, a knowledge graph-based method for analyzing the transmission of family credit risk determines the risk transmission data of the target family based on the target family's member knowledge graph, the individual risk value of each family member, risk transmission sub-data, and risk transmission matrix. This includes: inputting the target family's member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix into a risk coupling model to determine the risk coupling value of each family member; and determining the risk transmission data of the target family based on the individual risk values ​​and risk coupling values ​​of all family members.

[0012] This invention provides a knowledge graph-based family credit risk transmission analysis system for executing any one of the knowledge graph-based family credit risk transmission analysis methods in Examples 1 to 7. The system includes: a calculation module for acquiring family member data of a target family and calculating the individual risk value of each family member based on this data; a construction module for acquiring family transaction data of the target family and constructing a family member knowledge graph and a risk transmission matrix for the target family based on this data; a determination module for determining risk transmission sub-data for each family member based on the family member knowledge graph and the risk transmission matrix; and a transmission module for determining the risk transmission data of the target family based on the family member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix.

[0013] The beneficial effects of this invention compared to existing technologies are as follows: By acquiring data on family members and family interactions within a target family, calculating the individual risk value of each family member, constructing a knowledge graph of the target family members, and building a risk transmission matrix for the target family, this invention identifies the risk transmission sub-data for each family member and determines the overall risk transmission data for the target family. Based on this knowledge graph, a networked and systematic analysis of family credit risk can be achieved, accurately uncovering hidden kinship-related risks, improving the comprehensiveness and accuracy of risk prediction, providing high-value decision support for family credit risk management, and adapting to the risk analysis needs of complex family structures.

[0014] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 is a flowchart of a knowledge graph-based family credit risk transmission analysis method according to an embodiment of the invention; Figure 2 is a schematic diagram of a knowledge graph-based family credit risk transmission analysis system according to an embodiment of the invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention. Embodiment 1:

[0018] This invention provides a knowledge graph-based method for analyzing the transmission of family credit risk. Referring to Figure 1, the method includes: S1: acquiring family member data of the target family, and calculating the individual risk value of each family member based on the family member data; S2: acquiring family transaction data of the target family, constructing a family member knowledge graph of the target family based on the family member data and the family transaction data, and constructing a risk transmission matrix of the target family; S3: determining the risk transmission sub-data of each family member based on the family member knowledge graph and the risk transmission matrix; S4: determining the risk transmission data of the target family based on the family member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix.

[0019] In this embodiment, data on family members of the target family is acquired. This data includes each family member's name, generational ranking, and various personal credit data, such as personal credit score, personal assets, income and expenditure data for multiple specified time periods, and detailed historical delinquency data. Based on this comprehensive family member data, a multi-dimensional comprehensive assessment of each family member's credit status is conducted to calculate an individual risk value that reflects the individual's credit risk level. This individual risk value serves as the individual risk benchmark for subsequent analysis.

[0020] In this embodiment, family transaction data of the target family is acquired. This data records in detail the kinship relationships and various economic transactions between each pair of family members, including specific information on various transaction types such as joint loans, joint mortgages, guarantees, and daily cash flow, covering transaction amounts, time periods, and corresponding entity information. Using family members as the core entities and kinship relationships as the connecting links, attributes such as member seniority are integrated to construct a member knowledge graph that intuitively presents the network of relationships within the family. Based on the previously acquired family member data, the identity and relationships of each member are clarified, the family transaction data is quantified, and the risk transmission correlation strength between each pair of family members is analyzed, thereby constructing a risk transmission matrix for the target family.

[0021] In this embodiment, a constructed member knowledge graph and a risk transmission matrix are used as the core basis for mining and filtering risk transmission paths. The member knowledge graph provides a clear framework of the family network, clarifying the hierarchical relationships and connection paths among members, while the risk transmission matrix provides quantitative data on the intensity of risk transmission among members. Combining these two types of data, starting with each family member, potential risk transmission paths within the family network are mined. The validity of these paths is judged based on the transmission values ​​in the risk transmission matrix. After eliminating invalid paths, the risk transmission sub-data corresponding to each family member is determined.

[0022] In this embodiment, the member knowledge graph clarifies the overall relational structure, individual risk values ​​provide individual risk benchmarks, risk transmission sub-data identifies effective transmission paths, and the risk transmission matrix quantifies transmission intensity. Through collaborative analysis of these data, risk transmission data that comprehensively reflects the overall picture of credit risk transmission within the target family is determined.

[0023] The beneficial effects of the above technologies are as follows: By acquiring data on family members and family interactions of the target family, calculating the individual risk value of each family member, constructing a knowledge graph of the target family members, and building a risk transmission matrix for the target family, the risk transmission sub-data of each family member can be determined, and the overall risk transmission data of the target family can be identified. This allows for networked and systematic analysis of family credit risk based on the knowledge graph, accurately identifying hidden kinship-related risks, improving the comprehensiveness and accuracy of risk prediction, providing high-value decision support for family credit risk management, and adapting to the risk analysis needs of complex family structures. Example 2:

[0024] Based on Example 1, a knowledge graph-based family credit risk transmission analysis method is proposed to obtain family member data of a target family, including: obtaining the name, generational level, and personal credit data of each family member of the target family. The personal credit data includes personal credit score, personal assets, periodic income and liabilities over multiple specified time periods, and historical overdue data. The historical overdue data includes sub-data of multiple historical overdue periods, which includes overdue type tags, overdue start date, overdue end date, overdue amount, and overdue status tags. Overdue type tags include credit cards, consumer loans, mortgages, business loans, etc., and overdue status tags include settled, written off, and repaid. Based on the names, generational levels, and personal credit data of all family members of the target family, the family member data of the target family is determined.

[0025] In this embodiment, the specified time period is one month, and all specified time periods can be 12, 24, 36, etc.

[0026] In this embodiment, three core types of information are collected from each family member in the target family: member name, generational ranking, and personal credit data. Member name serves as the basic identifier distinguishing different family members. Generational ranking clarifies the vertical hierarchical position of family members within the family, clearly reflecting the generational connections between members. Personal credit data is the core basis for assessing the creditworthiness of family members; among them, personal credit scores are the most commonly used credit scores by banks, consumer finance companies, and internet finance companies, ranging from 350 to 950. Personal assets reflect a member's wealth accumulation and debt repayment capacity. Periodic income and periodic liabilities over multiple specified time periods comprehensively present a member's monthly economic income and expenditure. Historical overdue data records a member's past credit defaults in detail, including multiple historical overdue sub-data points, each containing rich and detailed information. Overdue type tags clearly identify the type of credit business involved in the overdue payment. The overdue start date and overdue end date define the time span of the overdue behavior. The overdue amount directly reflects the severity of the overdue behavior. The overdue status label reflects the current processing result of overdue transactions.

[0027] In this embodiment, after collecting the names, generational rankings, and personal credit data of all family members of the target family, the family member data of the target family is formed through system integration and structured processing.

[0028] The beneficial effects of the above technologies are: acquiring data on family members of the target family can provide comprehensive data support for calculating the individual risk value of each family member and constructing a knowledge graph of the target family members. Example 3:

[0029] Based on Example 2, a knowledge graph-based family credit risk transmission analysis method calculates the personal risk value of each family member of the target family based on the family member data of the target family. This includes: calculating the credit risk value, periodic risk value, asset risk value, and delinquency risk value of each family member of the target family based on the personal credit data of each family member in the family member data of the target family, and calculating the personal risk value of each family member of the target family.

[0030] In this embodiment, based on the personal credit data of each family member in the target family's family member data, the credit risk value, periodic risk value, asset risk value, and delinquency risk value of each family member in the target family are calculated, and the personal risk value of each family member in the target family is calculated. The calculation formula is expressed as follows: ;in, This represents the individual risk value of the i-th family member in the target family. , Let these represent the credit risk value, periodic risk value, asset risk value, and delinquency risk value of the i-th family member of the target family, respectively. This represents the personal credit score from the personal credit data of the i-th family member in the target family. This represents the personal assets of the i-th family member in the target family, as shown in their personal credit data. These represent the periodic income and periodic debt of the i-th family member in the target family during the k-th specified period, respectively. Let BS represent the average periodic income and average periodic debt for all specified periods in the personal credit data of the i-th family member of the target family, respectively, and BS represent the benchmark credit score. This represents the credit reporting scale adjustment factor, and N1 represents the number of specified periods. Let α represent the debt service ratio for all specified periods in the personal credit data of the i-th family member of the target household, and let α represent the disposable income adjustment factor. Let THA represent the periodic disposable income for all specified periods in the personal credit data of the i-th family member of the target household, γ represent the minimum asset threshold, β represent the asset penalty factor, and β represent the asset scaling factor. This represents the delinquency type label within the historical delinquency sub-data of the j-th historical delinquency in the personal credit data of the i-th family member of the target household. This represents the start date and end date of the overdue payment in the historical overdue data of the j-th historical overdue payment in the personal credit data of the i-th family member of the target family. This represents the overdue amount in the historical overdue sub-data of the j-th historical overdue payment in the historical overdue data of the personal credit data of the i-th family member of the target family. This represents the overdue status label in the historical overdue data of the j-th historical overdue instance within the historical overdue data of the i-th family member's personal credit data in the target family. This represents the state weight of the overdue status label in the historical overdue data of the j-th historical overdue instance within the historical overdue data of the i-th family member's personal credit data of the target family. Let represent the type weight of the delinquency type label in the historical delinquency sub-data of the j-th historical delinquency in the personal credit data of the i-th family member of the target family, where i ∈ N1 is the number of historical delinquencies in the historical delinquency data of the personal credit data of the i-th family member of the target family, and tc represents the current date. ρ1 represents the overdue amount scaling factor, and ρ2 represents the individual time decay factor. This represents the number of days between the start date and end date of the overdue payment in the historical overdue data of the j-th historical overdue payment in the personal credit data of the i-th family member of the target family.

[0031] In this embodiment, This represents the average disposable income over all specified periods in the personal credit data of the i-th family member of the target household. In this embodiment, 0.5 represents the debt service ratio safety threshold. This indicates that the debt service ratio is within a normal range and no penalty will be imposed. The more the risk exceeds the safety line, the more the risk index accelerates exponentially.

[0032] In this embodiment, the overdue amount scaling factor is set differently depending on the overdue type label. For example, if the overdue type label is credit card, the overdue amount scaling factor is... The amount can be 20,000, the overdue type is labeled as consumer loan, and the overdue amount scale factor is... The amount can be 50,000, the overdue type is labeled as business loan, and the overdue amount scale factor is... The amount can be 200,000, the overdue type is labeled as mortgage, and the overdue amount scale factor is... It can be 1,000,000.

[0033] In this embodiment, the smaller the asset scale factor β, the faster the risk is mitigated, and a small amount of assets can significantly reduce the risk. Conversely, the larger the asset scale factor β, the slower the risk is mitigated, and only a large amount of assets can significantly reduce the risk.

[0034] In this embodiment, the scale adjustment factor The value range is 50-150, and a possible value is 80. (Scale adjustment factor) The smaller the value, the more sensitive the relationship between an individual's credit score and its risk value; that is, a small change in the individual's credit score will result in a significant change in the risk value. (Scale Adjustment Factor) The higher the value, the less sensitive the credit score risk value is to changes in the individual's credit score. In other words, the credit score risk value will only change significantly when the individual's credit score changes considerably.

[0035] The beneficial effects of the above technology are as follows: Based on the family member data of the target family, the individual risk value of each family member can be calculated, enabling a refined assessment of individual risk. This provides high-quality data support for determining the risk transmission sub-data of each family member, and adapts to the risk analysis needs of complex family structures. Example 4:

[0036] Based on Example 1, a knowledge graph-based method for analyzing the transmission of family credit risk involves acquiring family transaction data of a target family. This includes: acquiring the kinship relationship and transaction data between every two family members of the target family. The kinship transaction data includes multiple sub-data of kinship transactions for all specified time periods, or may be empty. The sub-data of kinship transactions includes transaction type tags, transaction amounts, transaction start dates, and transaction end dates. Transaction type tags include joint lending, joint mortgage, guarantee, and daily turnover. Transaction type tags include the names of the guarantor and the guaranteed party when providing a guarantee, and the names of the transferor and the recipient when providing daily turnover. Based on the kinship transaction data between all family members of the target family, the family transaction data of the target family is determined.

[0037] In this embodiment, comprehensive basic information and specific details of interactions among members within the target family are collected. The kinship relationship between each pair of family members and the corresponding kinship interaction data are obtained. Kinship relationship is the foundation for defining the core attributes of relationships between family members, clarifying the kinship hierarchy and nature of the relationship. Kinship interaction data is crucial for reflecting the actual economic interactions and responsibility binding between members. Its coverage must include relevant data within all specified time periods to ensure comprehensive capture of interactions at different times, while also allowing for the existence of empty data to accommodate real-world scenarios where some family members have no interaction within a specified period. Each set of family transaction data consists of multiple sub-data sets, each containing a corresponding transaction type tag. These tags clarify the specific nature of each transaction, covering major forms of internal family economic relationships such as joint lending, joint mortgages, guarantees, and daily cash flow. Different transaction types correspond to different liability responsibilities. The transaction amount directly reflects the scale of the transaction; for example, when the transaction type tag is joint lending, joint mortgages, or guarantees, the transaction amount is the total amount of the corresponding business. The start and end dates of the transaction define the time span of each transaction, reflecting its continuity and time distribution characteristics. For different transaction type tags, specific key information needs to be collected to fully reconstruct the transaction scenario. When the transaction type tag is a guarantee, the names of the guarantor and the guaranteed party must be clearly recorded; this is core information for defining the attribution of guarantee liability and clarifying the subject of risk transmission. When the transaction type tag is daily cash flow, the names of the transferor and the recipient must be recorded to accurately locate the subject of the fund flow and clearly present the financial interaction relationship between members.

[0038] In this embodiment, the data on kinship exchanges among all family members of the target family are integrated to determine the family exchange data of the target family.

[0039] The beneficial effects of the above technologies are as follows: acquiring the family transaction data of the target family enables the full-cycle, full-scenario capture of internal family relationships, fully presenting the economic ties and responsibility networks among family members, and providing comprehensive and high-quality data support for in-depth exploration of family credit risk transmission paths. Example 5:

[0040] Based on Example 4, a knowledge graph-based family credit risk transmission analysis method is proposed. This method constructs a family member knowledge graph of the target family based on family member data and family transaction data, and then constructs a risk transmission matrix for the target family. The method includes: constructing the family member knowledge graph based on the member names, generational ranks, and kinship relationships between any two family members in the family transaction data; if the kinship transaction data between two family members in the target family's transaction data is not empty, calculating the transmission transaction amount in each kinship transaction sub-data in the kinship transaction data corresponding to each family member other than the stated family member, and calculating the risk transmission value of each family member to each family member other than the stated family member; if the kinship transaction data between two family members in the target family's transaction data is empty, determining the risk transmission value between the two family members of the stated family member to be 0; and constructing a risk transmission matrix for the target family based on the risk transmission values ​​among all family members of the target family.

[0041] In this embodiment, the core of constructing a member knowledge graph of the target family is to build a structured relational network based on three types of core information, strictly adhering to the row-by-row layout rules corresponding to generational levels. Member names are the core entity nodes of the graph; each member name is an independent node used to uniquely identify different individuals within the family. Generational levels are the basis for the graph's layout, explicitly stipulating that the family member nodes corresponding to the first generation are arranged in the first row, the family member nodes corresponding to the second generation are arranged in the second row, and so on, with subsequent generations arranged sequentially in rows, visually presenting the vertical hierarchical relationship within the family through row division. The kinship relationship between each pair of family members is the association edge connecting different nodes, used to link nodes in different rows to form a complete family relational network. Specifically, the names and corresponding generational levels of all members are first extracted from the family member data. Member name nodes are then divided into different rows according to generational level, with first-generation members concentrated in the first row and second-generation members concentrated in the second row, ensuring that members of the same generation are at the same level. Next, the kinship relationships between every two family members are extracted from the family interaction data, such as the parent-child relationship, the marital relationship between spouses, and the sibling relationship. These kinship relationships are used as connecting edges to link the corresponding member name nodes. For example, the grandparent member nodes in the first row are connected to the parent member nodes in the second row through the parent-child relationship edge. If there is a marital relationship between the parent member nodes in the second row, they are connected by the marital relationship edge. At the same time, the parent member nodes can also be connected to the child member nodes in the third row through the parent-child relationship edge. Finally, a visual member knowledge graph is formed, with nodes arranged in rows according to generation and kinship relationships as edges, clearly showing the hierarchical structure of family members and kinship relationships.

[0042] In this embodiment, based on the kinship data between every two family members in the multi-source family data of the target family, the risk transmission value of each family member of the target family to every family member other than the stated family member is calculated. The calculation formula is expressed as follows: ;in, Risk transmission value from the i-th family member to the p-th family member in the target family, where i ≠ p. This represents the risk transmission weight between the i-th family member and the p-th family member based on the relationship type label in the a-th sub-data of the corresponding kinship relationship data. This represents the amount of family transactions in the a-th sub-data of the family transaction data between the i-th and p-th family members of the target family. These represent the relationship type labels in the a-th sub-data of the relationship data of the i-th and p-th family members of the target family, respectively. This represents the start and end dates of the kinship transactions between the i-th and p-th family members of the target family, specifically within the a-th kinship transaction sub-data. This represents the number of months between the start date and end date of the relationship exchange in the a-th sub-data of the relationship exchange data between the i-th and p-th family members of the target family. This represents the number of months between the end date of the relationship and the current date in the a-th sub-data of the relationship data between the i-th and p-th family members of the target family. This represents the amount of money transferred between the i-th family member and the p-th family member based on the a-th sub-data of the corresponding kinship transaction data. This indicates the k-th specified period. Let ipN3 represent the risk transmission factor from the i-th family member of the target family to the p-th family member based on the a-th sub-data point of the corresponding kinship interaction data; let ipN3 represent the number of kinship interaction sub-data points in the kinship interaction data of the i-th and p-th family members of the target family; let ρ2 represent the interaction time decay factor; and let N4 represent the number of family members of the target family. This represents the risk time weight of the i-th family member of the target family to the p-th family member based on the a-th sub-data of the corresponding kinship interaction data. This represents the name of the i-th family member in the target family.

[0043] In this embodiment, for the scenario where there is no kinship interaction data between two family members in the family interaction data, i.e., the kinship interaction data is empty, there is no need for a complex calculation process; the risk transmission value between these two family members is directly and clearly defined as 0. The core basis of this definition logic is that when there is no kinship interaction between two family members, there is no direct economic bond between them, and therefore no carrier or path for risk transmission; thus, the possibility of risk transmission is 0.

[0044] In this embodiment, after calculating the risk transmission values ​​between all pairs of family members in the target family, a risk transmission matrix for the target family is constructed based on these calculation results. The core of the construction process is the systematic and structured integration of the dispersed risk transmission values ​​between pairs of members. The risk transmission matrix uses family members as the core dimension; the rows and columns of the matrix correspond to each family member in the target family, and each element in the matrix corresponds to the risk transmission value between the family member in its row and the family member in its column. This matrix-based presentation transforms the originally dispersed risk transmission relationships between pairs of members into a comprehensive and intuitive structured network.

[0045] The beneficial effects of the above technologies are as follows: Based on the target family's member data and family interaction data, a member knowledge graph of the target family is constructed, and a risk transmission matrix of the target family is built. This enables accurate quantification and comprehensive coverage of risk transmission values. The system presents the strength of risk transmission correlations among all family members, clearly outlining the internal risk transmission network of the family, and adapting to the risk analysis needs of families with different interaction densities. Example 6:

[0046] Based on Example 5, a knowledge graph-based family credit risk transmission analysis method is proposed. Based on the target family's member knowledge graph and risk transmission matrix, the method determines the risk transmission sub-data for each family member. This includes: based on the target family's member knowledge graph, starting with each family member, extending two generational levels vertically and horizontally along the same generational level to determine multiple risk transmission paths for each family member; if any two family members in a risk transmission path have a risk transmission value of 0, the path label for that family member's risk transmission path is determined to be discarded; if no two family members in a risk transmission path have a risk transmission value of 0, the path label for that family member's risk transmission path is determined to be normal; based on all risk transmission paths with normal path labels for each family member, the method determines the risk transmission sub-data for each family member.

[0047] In this embodiment, based on the target family's member knowledge graph, the risk transmission path corresponding to each family member is accurately mined. The mining process has clear scope constraints and directional guidance. Each family member is used as an independent starting point, fully utilizing the key attribute of generational level already marked in the member knowledge graph to conduct targeted path extension. The extension direction is clearly defined as two directions, upward and downward, and the corresponding extension range is strictly limited to two generational levels. Simultaneously, horizontal extension within the same generation is also carried out. This targeted and limited-range extension method effectively ensures that the mined risk transmission path focuses on the core kinship circle most closely related to the starting member. For example, starting with parents, extending upwards by two generational levels can cover grandparents and great-grandparents, and extending downwards by two generational levels can cover children and grandchildren. This comprehensively captures the core related nodes of the starting member in the vertical hierarchy of the family kinship network, avoiding excessively broad path coverage due to unlimited extension, which might include distant relatives or other irrelevant members, thus causing redundancy and interference in subsequent analysis. In this way, multiple potential risk transmission paths can be systematically identified for each family member. These paths comprehensively cover the direct and indirect connections between the starting member and family members of the two generations above and below, as well as among peers, clearly presenting the potential risk transmission channels within the family for the starting member.

[0048] In this embodiment, after mining each risk transmission path, the validity of each path needs to be rigorously determined. The core criterion for this determination is whether the risk transmission value between any two family members within the path is 0. The specific determination logic is divided into two cases. The first case is if any pair of family members in a risk transmission path has a risk transmission value of 0. This means that there is no effective economic connection or responsibility binding between these family members, and the basic conditions for risk transmission are not met. Therefore, the entire path loses its practical significance for risk transmission, and the path label for this risk transmission path is determined to be discarded. The second case is if the risk transmission value between all pairs of family members in a risk transmission path is not 0. This indicates that there is an effective relationship between all members on this path, possessing the carrier and possibility of risk transmission. This path is a valid path with practical analytical value, and therefore its path label is determined to be normal.

[0049] In this embodiment, after labeling all paths, the risk transmission paths corresponding to each family member are filtered and integrated, retaining only those with normal risk transmission labels. This targeted integration forms the risk transmission sub-data for each family member in the target family. The risk transmission sub-data centrally carries information on all valid risk transmission paths corresponding to each family member.

[0050] The beneficial effects of the above technologies are as follows: Based on the target family's member knowledge graph and risk transmission matrix, the risk transmission sub-data for each family member of the target family can be determined. This allows for focusing on the core kinship circle, avoiding interference from irrelevant paths, focusing on effective risk paths, and providing accurate and high-quality core data support for subsequent risk analysis. It also adapts to the risk transmission characteristics of hierarchical family relationships. Example 7:

[0051] Based on Example 1, a knowledge graph-based method for analyzing the transmission of family credit risk is proposed. This method determines the risk transmission data of the target family based on the target family's member knowledge graph, the individual risk value of each family member, risk transmission sub-data, and risk transmission matrix. The method includes: inputting the target family's member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix into a risk coupling model to determine the risk coupling value of each family member; and determining the risk transmission data of the target family based on the individual risk values ​​and risk coupling values ​​of all family members.

[0052] In this embodiment, a risk coupling model is used to achieve deep fusion of multi-source core data, thereby accurately quantifying the comprehensive risk status of each family member. Specifically, four types of key data need to be input into the risk coupling model. These four types of data each carry risk information of different dimensions, complementing each other to form a complete risk analysis data system. The member knowledge graph provides the network structure of family members within the target family and the generational attributes of each member, clearly defining the potential network framework of risk transmission and providing a structural foundation for the model to analyze the transmission path of risk. The personal risk value of each family member is a basic quantitative result of the individual's own credit risk level, covering multi-dimensional subdivided risk characteristics such as credit reporting cycle and asset delinquency, and is the core benchmark data for the model to assess the intensity of risk sources. The risk transmission sub-data focuses on the effective risk transmission path corresponding to each family member, eliminating paths without actual transmission significance, accurately locking the key channels of risk transmission for the model, and ensuring that the analysis is not interfered with by redundant information. The risk transmission matrix systematically presents the quantitative value of the risk transmission intensity between each pair of family members, providing a direct basis for the model to judge the possibility and intensity of risk transmission between different members. The core function of the risk coupling model is to break the isolation of these four types of data, conduct in-depth collaborative analysis, explore the interaction and superposition effect between individual risk and family-related risk, and finally output a risk coupling value for each family member that can comprehensively reflect their own risk level and the transmission effect of related risks. This value is no longer a single individual risk or an isolated related risk, but a comprehensive risk representation after the deep coupling of the two.

[0053] In this embodiment, after obtaining the individual risk values ​​and risk coupling values ​​of all family members in the target family, the risk transmission data of the target family is finally determined through the systematic integration and comprehensive analysis of these two types of core risk indicators. The individual risk value, as the basic benchmark for individual risk, clarifies the initial level and source of risk for each family member, serving as a key basis for identifying the starting point of risk transmission. The risk coupling value supplements the information by showing the cumulative impact of risk on each member after it is transmitted through related paths within the family, clearly reflecting the amplification or cumulative effect of related risks on individual risk. During the integrated analysis, it is necessary not only to clarify the risk level of each member but also to trace the transmission logic of risk within the family based on these two types of data. This includes key information such as which effective paths the risk is transmitted from its initial source, how the risk intensity changes during transmission, which members are the core nodes of risk transmission, and the extent of the risk's ultimate impact on the family as a whole. Through such systematic integration, the final risk transmission data can completely and comprehensively present the entire picture of credit risk transmission in the target family, including both individual-level risk details and the overall risk transmission network characteristics of the family.

[0054] In this embodiment, the core principle of the risk coupling model is the deep integration of risk-related data to achieve the synergistic quantification of individual risks among family members and related risks within the family. Based on the family hierarchy and kinship framework provided by the member knowledge graph, the model anchors the network boundaries and hierarchical constraints of risk transmission; it uses individual risk values ​​as the individual risk benchmark to clarify the initial risk level of each member; it uses effective transmission paths locked by risk transmission sub-data as the core analysis link to avoid interference from ineffective paths; and it uses the transmission strength between members quantified by the risk transmission matrix as the correlation weight to accurately measure the transmission effect of risk in the path. By exploring the interaction mechanism between individual risk and related risks, the model comprehensively calculates the superposition effect of individual risk and related risks transmitted through effective paths, ultimately outputting a risk coupling value that comprehensively reflects the overall risk status of each family member.

[0055] The beneficial effects of the above technologies are as follows: Based on the target family's member knowledge graph, each family member's individual risk value, risk transmission sub-data, and risk transmission matrix, the risk transmission data of the target family can be determined. This enables precise coupling and quantification of individual risk and family-related risk, providing high-dimensional and highly reliable data support for the overall assessment and precise control of family credit risk, and adapting to the risk analysis needs of complex family relationship networks. Example 8:

[0056] This invention provides a knowledge graph-based family credit risk transmission analysis system for executing any of the knowledge graph-based family credit risk transmission analysis methods in Examples 1 to 7. Referring to Figure 2, the system includes: a calculation module for acquiring family member data of the target family and calculating the individual risk value of each family member based on the family member data; a construction module for acquiring family transaction data of the target family and constructing a member knowledge graph of the target family and a risk transmission matrix of the target family based on the family member data and the family transaction data; a determination module for determining the risk transmission sub-data of each family member based on the member knowledge graph and the risk transmission matrix; and a transmission module for determining the risk transmission data of the target family based on the member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix.

[0057] The beneficial effects of the above technologies are as follows: By acquiring data on family members and family interactions within the target family, calculating the individual risk value of each family member, constructing a knowledge graph of the target family members, and building a risk transmission matrix for the target family, the technologies determine the risk transmission sub-data for each family member and the overall risk transmission data for the target family. This knowledge graph enables networked and systematic analysis of family credit risk, accurately uncovering hidden kinship-related risks, improving the comprehensiveness and accuracy of risk prediction, providing high-value decision support for family credit risk management, and adapting to the risk analysis needs of complex family structures.

[0058] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for analyzing the transmission of household credit risk based on knowledge graphs, characterized in that, include: S1: Obtain the family member data of the target family, and calculate the individual risk value of each family member based on the family member data of the target family; S2: Obtain the family interaction data of the target family, construct a member knowledge graph of the target family based on the family member data and family interaction data, and construct a risk transmission matrix of the target family; S3: Based on the target family's member knowledge graph and risk transmission matrix, determine the risk transmission sub-data for each family member in the target family; S4: Based on the target family's member knowledge graph, each family member's personal risk value, risk transmission sub-data, and risk transmission matrix, determine the target family's risk transmission data.

2. The method for analyzing the transmission of family credit risk based on knowledge graphs according to claim 1, characterized in that, Obtaining family member data for the target family includes: obtaining the name, generational ranking, and personal credit data of each family member. Personal credit data includes personal credit score, personal assets, periodic income and debt over multiple specified time periods, and historical delinquency data. Historical delinquency data includes multiple delinquency sub-data points, including delinquency type tags, delinquency start date, delinquency end date, delinquency amount, and delinquency status tags. Delinquency type tags include credit cards, consumer loans, mortgages, business loans, etc., and delinquency status tags include settled, written off, and repaid. Based on the names, generational ranking, and personal credit data of all family members in the target family, the family member data of the target family is determined.

3. The method for analyzing the transmission of family credit risk based on knowledge graphs according to claim 2, characterized in that, Based on the family member data of the target family, calculate the personal risk value of each family member of the target family, including: based on the personal credit data of each family member in the family member data of the target family, calculate the credit risk value, cycle risk value, asset risk value, and delinquency risk value of each family member of the target family, and calculate the personal risk value of each family member of the target family.

4. The method for analyzing the transmission of family credit risk based on knowledge graphs according to claim 1, characterized in that, Obtain the target family's family transaction data, including: obtaining the kinship relationship and transaction data between every two family members of the target family. The family transaction data includes multiple sub-data of family transactions for all specified time periods or is empty. The sub-data of family transactions includes transaction type tags, transaction amount, transaction start date, and transaction end date. Transaction type tags include joint loan, joint mortgage, guarantee, and daily turnover. Transaction type tags are the names of the guarantor and the guaranteed party when providing a guarantee, and the names of the transferor and the recipient when providing daily turnover. Based on the family transaction data between all family members of the target family, determine the target family's family transaction data.

5. The method for analyzing the transmission of family credit risk based on knowledge graphs according to claim 4, characterized in that, Based on the target family's member data and family interaction data, a member knowledge graph of the target family is constructed, and a risk transmission matrix of the target family is also constructed. This includes: constructing the member knowledge graph of the target family based on the member names and generational levels of each member in the member data and the kinship relationships between every two family members in the family interaction data; if the kinship interaction data between two family members in the target family's family interaction data is not empty, calculating the transmission amount of each family member to each kinship interaction sub-data in the kinship interaction data corresponding to each family member other than the stated family member, and calculating the risk transmission value of each family member to each family member other than the stated family member; if the kinship interaction data between two family members in the target family's family interaction data is empty, determining that the risk transmission value between the two family members of the stated family member is 0; and constructing the risk transmission matrix of the target family based on the risk transmission values ​​among all family members of the target family.

6. The method for analyzing the transmission of family credit risk based on knowledge graphs according to claim 5, characterized in that, Based on the target family's member knowledge graph and risk transmission matrix, the risk transmission sub-data for each family member of the target family is determined, including: based on the target family's member knowledge graph, starting from each family member, extending two generational levels vertically and horizontally along the same generational level of the member knowledge graph to determine multiple risk transmission paths for each family member of the target family; if any two family members in the risk transmission path of the target family member have a risk transmission value of 0, the path label for that family member's risk transmission path is determined to be discarded; if no two family members in the risk transmission path of the target family member have a risk transmission value of 0, the path label for that family member's risk transmission path is determined to be normal; based on all risk transmission paths of each family member of the target family whose path labels are normal, the risk transmission sub-data for each family member of the target family is determined.

7. The method for analyzing the transmission of family credit risk based on knowledge graphs according to claim 1, characterized in that, Based on the target family's member knowledge graph, each family member's individual risk value, risk transmission sub-data, and risk transmission matrix, the risk transmission data of the target family is determined, including: inputting the target family's member knowledge graph, each family member's individual risk value, risk transmission sub-data, and risk transmission matrix into the risk coupling model to determine the risk coupling value of each family member of the target family; and determining the risk transmission data of the target family based on the individual risk values ​​and risk coupling values ​​of all family members of the target family.

8. A knowledge graph-based family credit risk transmission analysis system, characterized in that, A method for performing a family credit risk transmission analysis based on a knowledge graph, as described in any one of claims 1 to 7, comprises: a calculation module for acquiring family member data of a target family and calculating the individual risk value of each family member based on the family member data; a construction module for acquiring family transaction data of the target family and constructing a family member knowledge graph and a risk transmission matrix of the target family based on the family member data and the family transaction data; a determination module for determining the risk transmission sub-data of each family member based on the family member knowledge graph and the risk transmission matrix; and a transmission module for determining the risk transmission data of the target family based on the family member knowledge graph, the individual risk value of each family member, the risk transmission sub-data, and the risk transmission matrix.

Images