A Personalized Financial Service Recommendation Method and System Based on Intelligent Large Model

Abstract

This invention discloses a personalized financial service recommendation method and system based on an intelligent large-scale model, belonging to the field of service recommendation technology. After authorization from the user's family, basic social attribute information of family members is collected, and a relationship structure graph and financial behavior path chain are constructed. Based on the family members' family identity and operation frequency, the financial behavior path chain is sequentially merged; a family financial behavior matrix is ​​constructed; the financial behavior intensity value of family members is calculated; the behavioral preference value of all family members under each service type is calculated; a financial service behavior preference vector for the user's family is constructed, and the similarity of behavioral preferences between different user families is calculated; a preset threshold is used to analyze and provide unified financial service recommendations for the user's family. This invention constructs a family financial profile through an intelligent large-scale model, achieving personalized service matching while improving the systematicness and accuracy of financial services, significantly enhancing the relevance of recommendations and user satisfaction.

Description

Technical Field

[0001] This invention relates to the field of service recommendation technology, specifically to a personalized financial service recommendation method and system based on an intelligent large model. Background Technology

[0002] With the continuous development of artificial intelligence and big data technologies, the FinTech field is rapidly moving towards a stage of intelligent and precise services. Particularly in personalized financial service recommendations, traditional recommendation mechanisms based on single user attributes or individual behaviors have gradually revealed problems such as information silos, low recommendation accuracy, and lack of scenario adaptability. In recent years, breakthroughs in semantic understanding and multi-dimensional data fusion by large-scale models have provided a technological foundation for introducing intelligent large-scale models into financial service recommendations. Meanwhile, as a crucial unit in financial decision-making, the overall financial behavior characteristics, preference patterns, and interaction paths among family members are considered key entry points for precise services. Therefore, mining the financial behavior structure within families and inferring their potential preferences through intelligent large-scale models has become a new trend for achieving efficient financial service recommendations.

[0003] While some financial recommendation systems have incorporated deep learning models to optimize recommendations, they generally suffer from issues such as coarse-grained modeling of user behavior, a lack of analytical capabilities to link financial behaviors among family members, and high dispersion in recommendation results across multi-user scenarios. Furthermore, recommendations are often based solely on individual historical records, lacking a systematic analysis of the overall family behavior path and failing to effectively capture behavioral similarities among user families. This results in a lack of aggregation and targeting in service recommendations. This is particularly problematic in scenarios involving shared financial accounts, joint decision-making, and joint asset ownership among family members, where such recommendation mechanisms, which ignore group relationships, clearly fail to meet users' actual needs. Summary of the Invention

[0004] The purpose of this invention is to provide a personalized financial service recommendation method and system based on an intelligent large model to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A personalized financial service recommendation method based on an intelligent large model includes the following steps: Step S1: After authorization from the user's family, collect basic social attribute information of family members and construct a relationship structure diagram of the user's family; Step S2: Based on the relationship structure diagram, obtain the behavioral trajectory data of family members on the financial service platform and construct a financial behavior path chain; based on the family member's family identity and operation frequency, sequentially merge the financial behavior path chain; Step S3: Based on the sequentially merged financial behavior path chain, construct a family financial behavior matrix; calculate the financial behavior intensity value of family members under the service type; based on the financial behavior intensity value, calculate the behavioral preference value of all family members under the service type; Step S4: Based on the behavioral preference value, construct a financial service behavior preference vector of the user's family and calculate the behavioral preference similarity between different user families; preset behavioral preference value threshold and behavioral preference similarity threshold, analyze and perform unified financial service recommendations for user families.

[0007] As a preferred embodiment of the personalized financial service recommendation method based on an intelligent large model described in this invention, after authorization from the user's family, basic social attribute information of family members is collected. This basic social attribute information includes the age, marital status, and family identity of the family members. The basic social attribute information of all family members in the user's family is then obtained to construct a relationship structure diagram of the user's family, as detailed below:

[0008] Each family member in the user's family is treated as a graph node in the relational structure graph, and the basic social attribute information of the family members is treated as the node attribute of the graph node.

[0009] The system obtains the address information, joint ownership record information of family members, and registration information of financial service platforms of the user's family, and inputs them into the intelligent big data model. Through the intelligent big data model, it obtains the kinship data between family members in the user's family, and uses the kinship data between family members as edges in the relationship structure graph to construct the edge connection structure of the graph.

[0010] As a preferred embodiment of the personalized financial service recommendation method based on intelligent big data model described in this invention, the behavioral trajectory data of family members in the user's family on the financial service platform is obtained based on the relationship structure diagram of the user's family. The behavioral trajectory data includes the operation frequency, transaction amount and service type of family members on the financial service platform, and the operation frequency is in monthly units.

[0011] Based on the operation frequency, transaction amount, and service type, a financial behavior path chain for family members is constructed. The financial behavior path chains of all family members in the user's family are obtained, and based on the family identity of the family members, the financial behavior path chains of all family members in the user's family are classified (the financial behavior path chains of these members are grouped according to their family identity, that is, "the path chains of all family decision-makers are put together", "the path chains of all children are put together", etc.). The classified financial behavior path chains are merged according to the operation frequency of the service type (each family member will have different service types in the path chain (such as loans, wealth management, and payments). Some service types are repeated in multiple members. The behavioral trajectories of all members on the "same service type" are extracted, and then integrated into a unified path according to the order of operation frequency).

[0012] As a preferred embodiment of the personalized financial service recommendation method based on an intelligent large model described in this invention, a family financial behavior matrix is ​​constructed based on the sequentially merged financial behavior path chain. The rows of the family financial behavior matrix represent service types, the columns represent family members within the user's family, and the elements of the family financial behavior matrix are the financial behavior intensity values ​​of the corresponding family members under the service type. The formula for calculating the financial behavior intensity value is as follows:

[0013]

[0014] Among them, S ij F represents the intensity of financial behavior of the i-th family member under the j-th service type. ij Let F represent the frequency of operation of the i-th family member under the j-th service type, min(F j ) represents the minimum operation frequency of the user family under the j-th service type, max(F j A represents the maximum operating frequency of the user's household under the j-th service type. ij Let A represent the transaction limit of the i-th family member under the j-th service type, min(A j ) represents the minimum transaction limit for the user family under the j-th service type, max(A j ) represents the maximum transaction amount of the user family under the j-th service type, and α and β represent the influencing factors related to operation frequency and transaction amount, respectively;

[0015] In this invention, the formula comprehensively considers both operation frequency and transaction amount to calculate the intensity value of financial behavior. Among these factors, The operating frequency F ijNormalize the data and map it to the {0,1} interval to reflect the position of the member’s operation frequency relative to the maximum and minimum operation frequency within the family under the j-th service type. Regarding transaction amount A ij The normalization process involves α and β, which are influencing factors used to adjust the relative importance of operation frequency and transaction amount in calculating the intensity value of financial behavior. These factors can be set according to the actual situation. For example, in some financial scenarios that emphasize transaction activity, α can be set larger.

[0016] The financial behavior intensity value calculated using this formula can accurately measure the activity level and transaction scale influence of each family member across different service types. This provides foundational data for subsequent calculations of the family's overall behavioral preference value, helping to precisely characterize the financial behavior features of family members. For example, when analyzing family financial management services, this value can be used to determine the degree of participation of members in different financial products.

[0017] Based on the financial behavior intensity value S of the i-th family member under the j-th service type ij Calculate the behavioral preference values ​​of all family members in the user's household for the j-th service type using the following formula:

[0018]

[0019] Among them, BPV j ω represents the behavioral preference value of all family members in the user's household for the j-th service type. i This represents a preset weighting factor related to family members, where I represents the total number of family members in the user's family.

[0020] As a preferred embodiment of the personalized financial service recommendation method based on intelligent large model described in this invention, the method is based on the behavioral preference values ​​(BPV) of all family members in the user's household for the j-th service type. j Construct the financial service behavior preference vector of the user household, and denote the financial service behavior preference vector of the a-th user household as VSP. a Among them, VSP a =(BPV) a,j |j∈[1,J]), where BPV a,j represents the behavioral preference values ​​of all family members in the family of user a for service type j, where J represents the total number of service types;

[0021] Financial service behavior preference vector VSP based on user household a a The similarity of behavioral preferences among different user households is calculated using the following formula:

[0022]

[0023] Among them, BPS (VSP) a VSP a+1 BPV represents the similarity of behavioral preferences between user family a and user family a+1. a+1,j This represents the behavioral preference values ​​of all family members in the (a+1)th user's household for the jth service type;

[0024] In this invention, the formula uses vector dot product and vector magnitude to measure the similarity of behavioral preferences between two user households (the a-th and the (a+1)-th). (Molecular) It is the sum of the corresponding products of the two families' behavioral preferences for service types J, reflecting the consistency of the two families' preferences for each service type; the denominator is the sum of the corresponding products of the ... This involves normalizing the behavioral preference vectors of the two families so that the similarity calculation results are in the {0,1} interval, making comparison easier.

[0025] By calculating the similarity of behavioral preferences, financial institutions can identify family groups with similar behavioral preferences. For families with similar behavioral preferences, existing successful recommendation cases or service models can be referenced to improve the accuracy and effectiveness of recommendations and enhance user satisfaction. For example, if two families are found to have highly similar behavioral preferences, and one family shows a high acceptance of a newly launched financial product, the product can be recommended to the other family.

[0026] Preset behavioral preference value thresholds and behavioral preference similarity thresholds. If the behavioral preference value (BPV) of all family members in user a's household for service type j is... a,j If the behavioral preference value is greater than or equal to the threshold value, and the similarity of the behavioral preferences between the user family and the (a+1)th user family is greater than or equal to the threshold value, then it is determined that the (a)th user family has a behavioral preference for the (j)th service type and has a similar behavioral preference to the (a+1)th user family.

[0027] Obtain all service types with behavioral preferences in user family a and all family users with similar behavioral preferences to user family a. Based on all service types with behavioral preferences, make unified financial service recommendations to user families.

[0028] A personalized financial service recommendation system based on an intelligent large model, comprising: a data acquisition and graph construction module, a behavior trajectory and path chain construction module, a behavior matrix and preference value calculation module, and a similarity calculation, analysis, and recommendation module;

[0029] The data acquisition and graph construction module: after authorization from the user's family, collects basic social attribute information of family members in the user's family and constructs a relationship structure graph of the user's family;

[0030] The behavior trajectory and path chain construction module: Based on the relationship structure graph, it obtains the behavior trajectory data of family members on the financial service platform and constructs the financial behavior path chain; based on the family member's family identity and operation frequency, it sequentially merges the financial behavior path chain.

[0031] The behavior matrix and preference value calculation module: constructs a family financial behavior matrix based on the sequentially merged financial behavior path chain; calculates the financial behavior intensity value of family members under the service type; and calculates the behavior preference value of all family members under the service type based on the financial behavior intensity value.

[0032] The similarity calculation and analysis recommendation module: Based on behavioral preference values, it constructs a financial service behavior preference vector for user households and calculates the similarity of behavioral preferences between different user households; it presets behavioral preference value thresholds and behavioral preference similarity thresholds, analyzes and provides unified financial service recommendations for user households.

[0033] Furthermore, the data acquisition and graph construction module includes a data acquisition unit and a graph construction unit;

[0034] The data collection unit: after authorization from the user's family, collects basic social attribute information of the family members in the user's family, including the age, marital status and family identity of the family members;

[0035] The graph construction unit: It takes the basic social attribute information of all family members in the user's family and constructs a relationship structure graph of the user's family. Specifically, it uses each family member in the user's family as a graph node in the relationship structure graph, and uses the basic social attribute information of the family members as the node attributes of the graph nodes; it obtains the address information, family property co-ownership record information, and financial service platform registration information of the family members in the user's family, and inputs them into a smart big data model. Through the smart big data model, it obtains the kinship data between family members in the user's family, and uses the kinship data between family members as edges in the relationship structure graph to construct the edge connection structure of the graph.

[0036] Furthermore, the behavior trajectory and path chain construction module includes a behavior trajectory acquisition unit and a path chain construction unit;

[0037] The behavior trajectory acquisition unit: Based on the relationship structure diagram of the user's family, acquires the behavior trajectory data of family members in the user's family on the financial service platform. The behavior trajectory data includes the operation frequency, transaction amount and service type of family members on the financial service platform. The operation frequency is in monthly units.

[0038] The path chain construction unit: constructs a financial behavior path chain for family members based on the operation frequency, transaction amount, and service type; obtains the financial behavior path chains of all family members in the user's family, and classifies the financial behavior path chains of all family members in the user's family based on the family identity of the family members, and merges the classified financial behavior path chains in order of operation frequency according to the service type.

[0039] Furthermore, the behavior matrix and preference value calculation module includes a behavior matrix construction unit and a preference value calculation unit;

[0040] The behavior matrix construction unit: Based on the sequentially merged financial behavior path chain, it constructs a family financial behavior matrix. The rows of the family financial behavior matrix represent service types, the columns of the family financial behavior matrix represent family members in the user's family, and the elements of the family financial behavior matrix are the financial behavior intensity values ​​of the corresponding family members under the service type.

[0041] The preference value calculation unit calculates the behavioral preference values ​​of all family members in the user's family for the service type based on the intensity values ​​of their financial behavior under the service type.

[0042] Furthermore, the similarity calculation and analysis recommendation module includes a similarity calculation unit and an analysis and recommendation unit;

[0043] The similarity calculation unit: constructs a financial service behavior preference vector for the user's family based on the behavioral preference values ​​of all family members in the user's family for service types; and calculates the behavioral preference similarity between different user families based on the financial service behavior preference vector of the user's family.

[0044] The analysis and recommendation unit: presets a behavioral preference value threshold and a behavioral preference similarity threshold. If the behavioral preference values ​​of all family members in the a-th user's household for a service type are greater than or equal to the behavioral preference value threshold, and the behavioral preference similarity with the (a+1)-th user's household is greater than or equal to the behavioral preference similarity threshold, then it is determined that the a-th user's household has a behavioral preference for the j-th service type and has a similar behavioral preference with the (a+1)-th user's household. It obtains all service types for which the a-th user's household has a behavioral preference and all family users with similar behavioral preferences to the a-th user's household. Based on all service types for which the a-th user's household has a behavioral preference, it makes unified financial service recommendations to the user's household.

[0045] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: This invention provides a personalized financial service recommendation method and system based on an intelligent large-scale model. By collecting basic social attribute information of family members and constructing a relationship structure diagram of the user's family, it achieves a structured expression of the roles and relationships of family members, providing a data foundation and family semantic background for subsequent personalized modeling. Based on the relationship structure diagram, it acquires and integrates the behavioral trajectory data of family members on the financial service platform, classifies and merges path chains according to family identity, and achieves unified modeling of multi-member financial behavior, effectively revealing the collaborative characteristics of family financial decision-making. Based on the merged path chains, it constructs a financial behavior matrix, quantifies the behavioral intensity of each member in various services, and calculates behavioral preference values ​​accordingly, thereby achieving accurate characterization of individual behavior and collective preferences within the family. By constructing a family preference vector and calculating behavioral preference similarity, combined with threshold filtering of similar families and service types, it realizes a cross-family intelligent service recommendation mechanism. This invention breaks through the limitations of traditional individual-centered financial recommendations and utilizes an intelligent large-scale model to construct a structured, multi-dimensional family financial profile. While achieving personalized service matching, it improves the systematicness and accuracy of financial services, significantly enhancing the relevance, practicality, and user satisfaction of recommendations. Attached Figure Description

[0046] 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.

[0047] Figure 1 This is a schematic diagram illustrating the steps of a personalized financial service recommendation method based on an intelligent large model according to the present invention.

[0048] Figure 2 This is a schematic diagram of the structure of a personalized financial service recommendation system based on an intelligent large model according to the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please see Figure 1 In this first embodiment, a personalized financial service recommendation method based on an intelligent large model is provided. The method includes the following steps:

[0051] Step S1: After obtaining authorization from the user's family, collect basic social attribute information of family members in the user's family and construct a relationship structure diagram of the user's family.

[0052] Specifically, after obtaining authorization from the user's family, basic social attribute information of family members is collected, including the age, marital status, and family identity of the family members. Basic social attribute information of all family members is obtained, and a relationship structure diagram of the user's family is constructed, as follows:

[0053] Each family member in the user's family is treated as a graph node in the relational structure graph, and the basic social attribute information of the family members is treated as the node attribute of the graph node.

[0054] The system obtains the address information, joint ownership record information of family members, and registration information of financial service platforms of the user's family, and inputs them into the intelligent big data model. Through the intelligent big data model, it obtains the kinship data between family members in the user's family, and uses the kinship data between family members as edges in the relationship structure graph to construct the edge connection structure of the graph.

[0055] Step S2: Based on the relationship structure diagram, obtain the behavioral trajectory data of family members on the financial service platform and construct the financial behavior path chain; based on the family member's family identity and operation frequency, sequentially merge the financial behavior path chain.

[0056] Specifically, based on the relationship structure diagram of the user's family, the behavioral trajectory data of family members in the user's family on the financial service platform is obtained. The behavioral trajectory data includes the operation frequency, transaction amount and service type of family members on the financial service platform, and the operation frequency is in monthly units.

[0057] Furthermore, based on the operation frequency, transaction amount, and service type, a financial behavior path chain for family members is constructed; the financial behavior path chains of all family members in the user's family are obtained, and based on the family identity of the family members, the financial behavior path chains of all family members in the user's family are classified (the financial behavior path chains of these members are grouped according to their family identity, that is, "the path chains of all family decision-makers are put together", "the path chains of all children are put together", etc.), and the classified financial behavior path chains are merged according to the service type in order of operation frequency (each family member will have different service types in the path chain (such as loans, wealth management, payments), some service types appear repeatedly in multiple members, the behavioral trajectories of all members on "the same service type" are extracted, and then integrated into a unified path in order of operation frequency).

[0058] Step S3: Construct a family financial behavior matrix based on the sequentially merged financial behavior path chain; calculate the financial behavior intensity value of family members under the service type; and calculate the behavior preference value of all family members under the service type based on the financial behavior intensity value.

[0059] Specifically, based on the sequentially merged financial behavior path chain, a family financial behavior matrix is ​​constructed. The rows of the family financial behavior matrix represent service types, the columns represent family members within the user's family, and the elements of the family financial behavior matrix are the financial behavior intensity values ​​of the corresponding family members under each service type. The formula for calculating the financial behavior intensity value is as follows:

[0060]

[0061] Among them, S ij F represents the intensity of financial behavior of the i-th family member under the j-th service type. ij Let F represent the frequency of operation of the i-th family member under the j-th service type, min(F j ) represents the minimum operation frequency of the user family under the j-th service type, max(F j A represents the maximum operating frequency of the user's household under the j-th service type. ij Let A represent the transaction limit of the i-th family member under the j-th service type, min(A j ) represents the minimum transaction limit for the user family under the j-th service type, max(A j ) represents the maximum transaction amount of the user family under the j-th service type, and α and β represent the influencing factors related to operation frequency and transaction amount, respectively;

[0062] It's important to note that the formula normalizes the frequency of operations and transaction amounts to highlight the relative intensity of each family member's behavior within a specific service type within the family. For example, in a family of five, the father might operate 10 times per month for savings, with the family's minimum operation frequency being 2 times per month and the maximum being 15 times. After normalization, the formula clearly shows the father's relative position within the family, allowing for a fair comparison of each member's activity level across different service types, even with significant differences in operation frequency and transaction amount across different service types. α and β can be flexibly adjusted based on the financial service scenario and business focus. In credit card promotion scenarios, banks focus more on customer usage frequency, thus increasing the value of α to give operation frequency a greater weight in the calculation of financial behavior intensity, more accurately measuring member participation in credit card services, and providing data support for targeted marketing.

[0063] Furthermore, based on the financial behavior intensity value S of the i-th family member under the j-th service type... ij Calculate the behavioral preference values ​​of all family members in the user's household for the j-th service type using the following formula:

[0064]

[0065] Among them, BPV j ω represents the behavioral preference value of all family members in the user's household for the j-th service type. i This represents a preset weighting factor related to family members, where I represents the total number of family members in the user's family.

[0066] It should be noted that calculating behavioral preference values ​​by weighted summation of the financial behavior intensity values ​​of family members comprehensively considers the different roles and influences of different family members in financial decision-making. In family investment decisions, parents, as the main economic pillars, may have a higher weight in stock investments, and their behavioral intensity has a greater impact on the family's overall stock investment preference; while children, although less involved, still have a certain influence. The weighting factor can reasonably reflect this difference, resulting in financial service preferences that are more in line with the family's actual situation.

[0067] This formula comprehensively reflects a family's preference for specific types of financial services. Financial institutions can use this value to accurately understand each family's level of preference for different financial products or services. If a family has a high preference for insurance services, the financial institution can recommend suitable insurance packages tailored to that family's protection needs, improving the accuracy and relevance of the recommendations.

[0068] Step S4: Based on the behavioral preference values, construct the financial service behavioral preference vector of user households and calculate the behavioral preference similarity between different user households; preset the behavioral preference value threshold and the behavioral preference similarity threshold, analyze and make unified financial service recommendations for user households.

[0069] Specifically, based on the behavioral preference values ​​(BPV) of all family members in the user's household for the j-th service type. j Construct the financial service behavior preference vector of the user household, and denote the financial service behavior preference vector of the a-th user household as VSP. a Among them, VSP a =(BPV) a,j |j∈[1,J]), where BPV a,j represents the behavioral preference values ​​of all family members in the family of user a for service type j, where J represents the total number of service types;

[0070] Financial service behavior preference vector VSP based on user household aa The similarity of behavioral preferences among different user households is calculated using the following formula:

[0071]

[0072] Among them, BPS (VSP) a VSP a+1 BPV represents the similarity of behavioral preferences between user family a and user family a+1. a+1,j This represents the behavioral preference values ​​of all family members in the (a+1)th user's household for the jth service type;

[0073] It's important to note that using vector calculation to comprehensively and accurately measure the similarity between households based on behavioral preference values ​​across multiple service types is a robust approach. For example, consider two households: one with behavioral preference values ​​of [0.8, 0.6, 0.4] for financial management, loans, and insurance services, and the other with [0.7, 0.5, 0.3]. The similarity calculated using the formula reflects the degree of similarity between these two households across various financial service preferences, avoiding the limitations of measuring based on a single service type. Using this calculated behavioral preference similarity, financial institutions can group households with similar preferences together. For newly launched financial products, if they perform well in a particular household group, they can be recommended to other similar households, increasing the success rate and efficiency of recommendations, reducing indiscriminate recommendations, improving the utilization efficiency of financial service resources, and simultaneously enhancing user satisfaction and acceptance of financial services.

[0074] Furthermore, preset behavioral preference value thresholds and behavioral preference similarity thresholds are defined. If the behavioral preference value (BPV) of all family members in the a-th user's household for the j-th service type is... a,j If the behavioral preference value is greater than or equal to the threshold value, and the similarity of the behavioral preferences between the user family and the (a+1)th user family is greater than or equal to the threshold value, then it is determined that the (a)th user family has a behavioral preference for the (j)th service type and has a similar behavioral preference to the (a+1)th user family.

[0075] Obtain all service types with behavioral preferences in user family a and all family users with similar behavioral preferences to user family a. Based on all service types with behavioral preferences, make unified financial service recommendations to user families.

[0076] Please see Figure 2 In this second embodiment: a personalized financial service recommendation system based on intelligent large model is provided. The system includes: a data acquisition and graph construction module, a behavior trajectory and path chain construction module, a behavior matrix and preference value calculation module, and a similarity calculation and analysis recommendation module.

[0077] The data acquisition and graph construction module: after authorization from the user's family, collects basic social attribute information of family members in the user's family and constructs a relationship structure graph of the user's family;

[0078] The behavior trajectory and path chain construction module: Based on the relationship structure graph, it obtains the behavior trajectory data of family members on the financial service platform and constructs the financial behavior path chain; based on the family member's family identity and operation frequency, it sequentially merges the financial behavior path chain.

[0079] The behavior matrix and preference value calculation module: constructs a family financial behavior matrix based on the sequentially merged financial behavior path chain; calculates the financial behavior intensity value of family members under the service type; and calculates the behavior preference value of all family members under the service type based on the financial behavior intensity value.

[0080] The similarity calculation and analysis recommendation module: Based on behavioral preference values, it constructs a financial service behavior preference vector for user households and calculates the similarity of behavioral preferences between different user households; it presets behavioral preference value thresholds and behavioral preference similarity thresholds, analyzes and provides unified financial service recommendations for user households.

[0081] Furthermore, the data acquisition and graph construction module includes a data acquisition unit and a graph construction unit;

[0082] The data collection unit: after authorization from the user's family, collects basic social attribute information of the family members in the user's family, including the age, marital status and family identity of the family members;

[0083] The graph construction unit: It takes the basic social attribute information of all family members in the user's family and constructs a relationship structure graph of the user's family. Specifically, it uses each family member in the user's family as a graph node in the relationship structure graph, and uses the basic social attribute information of the family members as the node attributes of the graph nodes; it obtains the address information, family property co-ownership record information, and financial service platform registration information of the family members in the user's family, and inputs them into a smart big data model. Through the smart big data model, it obtains the kinship data between family members in the user's family, and uses the kinship data between family members as edges in the relationship structure graph to construct the edge connection structure of the graph.

[0084] Furthermore, the behavior trajectory and path chain construction module includes a behavior trajectory acquisition unit and a path chain construction unit;

[0085] The behavior trajectory acquisition unit: Based on the relationship structure diagram of the user's family, acquires the behavior trajectory data of family members in the user's family on the financial service platform. The behavior trajectory data includes the operation frequency, transaction amount and service type of family members on the financial service platform. The operation frequency is in monthly units.

[0086] The path chain construction unit: constructs a financial behavior path chain for family members based on the operation frequency, transaction amount, and service type; obtains the financial behavior path chains of all family members in the user's family, and classifies the financial behavior path chains of all family members in the user's family based on the family identity of the family members, and merges the classified financial behavior path chains in order of operation frequency according to the service type.

[0087] Furthermore, the behavior matrix and preference value calculation module includes a behavior matrix construction unit and a preference value calculation unit;

[0088] The behavior matrix construction unit: Based on the sequentially merged financial behavior path chain, it constructs a family financial behavior matrix. The rows of the family financial behavior matrix represent service types, the columns of the family financial behavior matrix represent family members in the user's family, and the elements of the family financial behavior matrix are the financial behavior intensity values ​​of the corresponding family members under the service type.

[0089] The preference value calculation unit calculates the behavioral preference values ​​of all family members in the user's family for the service type based on the intensity values ​​of their financial behavior under the service type.

[0090] Furthermore, the similarity calculation and analysis recommendation module includes a similarity calculation unit and an analysis and recommendation unit;

[0091] The similarity calculation unit: constructs a financial service behavior preference vector for the user's family based on the behavioral preference values ​​of all family members in the user's family for service types; and calculates the behavioral preference similarity between different user families based on the financial service behavior preference vector of the user's family.

[0092] The analysis and recommendation unit: presets a behavioral preference value threshold and a behavioral preference similarity threshold. If the behavioral preference values ​​of all family members in the a-th user's household for a service type are greater than or equal to the behavioral preference value threshold, and the behavioral preference similarity with the (a+1)-th user's household is greater than or equal to the behavioral preference similarity threshold, then it is determined that the a-th user's household has a behavioral preference for the j-th service type and has a similar behavioral preference with the (a+1)-th user's household. It obtains all service types for which the a-th user's household has a behavioral preference and all family users with similar behavioral preferences to the a-th user's household. Based on all service types for which the a-th user's household has a behavioral preference, it makes unified financial service recommendations to the user's household.

[0093] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0094] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A personalized financial service recommendation method based on an intelligent large model, characterized in that, The method includes the following steps: Step S1: After obtaining authorization from the user's family, collect basic social attribute information of family members in the user's family and construct a relationship structure diagram of the user's family; Step S2: Based on the relationship structure diagram, obtain the behavioral trajectory data of family members on the financial service platform and construct the financial behavior path chain; based on the family member's family identity and operation frequency, sequentially merge the financial behavior path chain; Step S3: Construct a family financial behavior matrix based on the sequentially merged financial behavior path chain; calculate the financial behavior intensity value of family members under the service type; and calculate the behavior preference value of all family members under the service type based on the financial behavior intensity value. Step S4: Based on behavioral preference values, construct financial service behavioral preference vectors for user households and calculate the similarity of behavioral preferences between different user households; preset behavioral preference value thresholds and behavioral preference similarity thresholds, analyze and provide unified financial service recommendations for user households; The specific implementation process of step S3 includes: Based on the sequentially merged financial behavior path chain, a family financial behavior matrix is ​​constructed. The rows of the family financial behavior matrix represent service types, the columns represent family members within the user's family, and the elements of the family financial behavior matrix are the financial behavior intensity values ​​of the corresponding family members under each service type. The formula for calculating the financial behavior intensity value is as follows: ； in, This represents the intensity of the financial behavior of the i-th family member under the j-th service type. This represents the frequency of operation by the i-th family member under the j-th service type. This represents the minimum operating frequency of the user's household under the j-th service type. This represents the maximum frequency of operation for the user's household under the j-th service type. This represents the transaction limit of the i-th family member under the j-th service type. This represents the minimum transaction limit for the user's family under the j-th service type. This represents the maximum transaction limit for the user's family under the j-th service type. and These represent the influencing factors related to operation frequency and transaction amount, respectively. Based on the intensity value of the financial behavior of the i-th family member under the j-th service type Calculate the behavioral preference values ​​of all family members in the user's household for the j-th service type using the following formula: ； in, This represents the behavioral preference value of all family members in the user's household for the j-th service type. This represents a preset weighting factor related to family members, where I represents the total number of family members in the user's family; The specific implementation process of step S4 includes: Based on the behavioral preference values ​​of all family members in the user's household for the j-th service type. Construct the financial service behavior preference vector of the user household, and denote the financial service behavior preference vector of the a-th user household as . ,in, ,in, represents the behavioral preference values ​​of all family members in the family of user a for service type j, where J represents the total number of service types; Financial service behavior preference vector based on the a-th user's household The similarity of behavioral preferences among different user households is calculated using the following formula: ； in, This represents the similarity of behavioral preferences between user family a and user family a+1. This represents the behavioral preference values ​​of all family members in the (a+1)th user's household for the jth service type; Preset behavioral preference value thresholds and behavioral preference similarity thresholds. If the behavioral preference values ​​of all family members in the a-th user's household for the j-th service type are... If the behavioral preference value is greater than or equal to the threshold value, and the similarity of the behavioral preferences between the user family and the (a+1)th user family is greater than or equal to the threshold value, then it is determined that the (a)th user family has a behavioral preference for the (j)th service type and has a similar behavioral preference to the (a+1)th user family. Obtain all service types with behavioral preferences in user family a and all family users with similar behavioral preferences to user family a. Based on all service types with behavioral preferences, make unified financial service recommendations to user families.

2. The personalized financial service recommendation method based on an intelligent large model according to claim 1, characterized in that, The specific implementation process of step S1 includes: After obtaining authorization from the user's family, basic social attribute information of family members is collected, including the age, marital status, and family identity of the family members. Basic social attribute information of all family members is obtained, and a relationship structure diagram of the user's family is constructed, as follows: Each family member in the user's family is treated as a graph node in the relational structure graph, and the basic social attribute information of the family members is treated as the node attribute of the graph node. The system obtains the address information, joint ownership record information of family members, and registration information of financial service platforms of the user's family, and inputs them into the intelligent big data model. Through the intelligent big data model, it obtains the kinship data between family members in the user's family, and uses the kinship data between family members as edges in the relationship structure graph to construct the edge connection structure of the graph.

3. The personalized financial service recommendation method based on an intelligent large model according to claim 2, characterized in that, The specific implementation process of step S2 includes: Based on the relationship structure diagram of the user's family, the behavioral trajectory data of family members in the user's family on the financial service platform is obtained. The behavioral trajectory data includes the operation frequency, transaction amount and service type of family members on the financial service platform. The operation frequency is in monthly units. Based on the operation frequency, transaction amount, and service type, a financial behavior path chain for family members is constructed; the financial behavior path chains of all family members in the user's family are obtained, and based on the family member's family identity, the financial behavior path chains of all family members in the user's family are classified, and the classified financial behavior path chains are merged in order of operation frequency according to the service type.

4. A personalized financial service recommendation system based on an intelligent large-scale model, executing the personalized financial service recommendation method based on an intelligent large-scale model as described in any one of claims 1-3, characterized in that, The system includes: a data acquisition and graph construction module, a behavior trajectory and path chain construction module, a behavior matrix and preference value calculation module, and a similarity calculation, analysis, and recommendation module; The data acquisition and graph construction module: after authorization from the user's family, collects basic social attribute information of family members in the user's family and constructs a relationship structure graph of the user's family; The behavior trajectory and path chain construction module: Based on the relationship structure graph, it obtains the behavior trajectory data of family members on the financial service platform and constructs the financial behavior path chain; based on the family member's family identity and operation frequency, it sequentially merges the financial behavior path chain. The behavior matrix and preference value calculation module: constructs a family financial behavior matrix based on the sequentially merged financial behavior path chain; calculates the financial behavior intensity value of family members under the service type; and calculates the behavior preference value of all family members under the service type based on the financial behavior intensity value. The similarity calculation and analysis recommendation module: Based on behavioral preference values, it constructs a financial service behavior preference vector for user households and calculates the similarity of behavioral preferences between different user households; it presets behavioral preference value thresholds and behavioral preference similarity thresholds, analyzes and provides unified financial service recommendations for user households.

5. The personalized financial service recommendation system based on an intelligent large model according to claim 4, characterized in that: The data acquisition and graph construction module includes a data acquisition unit and a graph construction unit; The data collection unit: after authorization from the user's family, collects basic social attribute information of the family members in the user's family, including the age, marital status and family identity of the family members; The graph construction unit: It takes the basic social attribute information of all family members in the user's family and constructs a relationship structure graph of the user's family. Specifically, it uses each family member in the user's family as a graph node in the relationship structure graph, and uses the basic social attribute information of the family members as the node attributes of the graph nodes; it obtains the address information, family property co-ownership record information, and financial service platform registration information of the family members in the user's family, and inputs them into a smart big data model. Through the smart big data model, it obtains the kinship data between family members in the user's family, and uses the kinship data between family members as edges in the relationship structure graph to construct the edge connection structure of the graph.

6. The personalized financial service recommendation system based on an intelligent large model according to claim 5, characterized in that: The behavior trajectory and path chain construction module includes a behavior trajectory acquisition unit and a path chain construction unit; The behavior trajectory acquisition unit: Based on the relationship structure diagram of the user's family, acquires the behavior trajectory data of family members in the user's family on the financial service platform. The behavior trajectory data includes the operation frequency, transaction amount and service type of family members on the financial service platform. The operation frequency is in monthly units. The path chain construction unit: constructs a financial behavior path chain for family members based on the operation frequency, transaction amount, and service type; obtains the financial behavior path chains of all family members in the user's family, and classifies the financial behavior path chains of all family members in the user's family based on the family identity of the family members, and merges the classified financial behavior path chains in order of operation frequency according to the service type.

7. A personalized financial service recommendation system based on an intelligent large model according to claim 6, characterized in that: The behavior matrix and preference value calculation module includes a behavior matrix construction unit and a preference value calculation unit; The behavior matrix construction unit: Based on the sequentially merged financial behavior path chain, it constructs a family financial behavior matrix. The rows of the family financial behavior matrix represent service types, the columns of the family financial behavior matrix represent family members in the user's family, and the elements of the family financial behavior matrix are the financial behavior intensity values ​​of the corresponding family members under the service type. The preference value calculation unit calculates the behavioral preference values ​​of all family members in the user's family for the service type based on the intensity values ​​of their financial behavior under the service type.

8. The personalized financial service recommendation system based on an intelligent large model according to claim 7, characterized in that: The similarity calculation and analysis recommendation module includes a similarity calculation unit and an analysis recommendation unit; The similarity calculation unit: constructs a financial service behavior preference vector for the user's family based on the behavioral preference values ​​of all family members in the user's family for service types; and calculates the behavioral preference similarity between different user families based on the financial service behavior preference vector of the user's family. The analysis and recommendation unit: presets a behavioral preference value threshold and a behavioral preference similarity threshold. If the behavioral preference values ​​of all family members in the a-th user's family for the service type are greater than or equal to the behavioral preference value threshold, and the behavioral preference similarity with the (a+1)-th user's family is greater than or equal to the behavioral preference similarity threshold, then it is determined that the a-th user's family has a behavioral preference for the j-th service type and has a similar behavioral preference with the (a+1)-th user's family. Obtain all service types with behavioral preferences in user family a and all family users with similar behavioral preferences to user family a. Based on all service types with behavioral preferences, make unified financial service recommendations to user families.

Images