Agriculture-related enterprise financial product pushing method, device and system

Abstract

The application provides an agricultural enterprise financial product pushing method, device and system, which can be used in the field of artificial intelligence technology. The method comprises the following steps: receiving model parameters transmitted by a plurality of enterprise terminals, wherein the model parameters are model parameters of an enterprise model obtained by training a preset model according to enterprise internal data by each enterprise terminal; obtaining joint model parameters of the preset model according to the model parameters of the plurality of enterprise terminals, and obtaining an agricultural identification model according to the joint model parameters and the preset model; obtaining an agricultural identification result by predicting a user through the agricultural identification model, obtaining a user discount success rate through a preset financial discount prediction model according to the agricultural identification result and user data, and determining pushing information corresponding to the discount success rate to be pushed to a user terminal. The application can improve the accuracy of agricultural enterprise identification and improve the success rate of financial product transactions.

Description

Methods, devices and systems for promoting financial products to agricultural enterprises Technical Field

[0001] This invention relates to the field of financial technology, particularly to the field of artificial intelligence, and especially to a method, apparatus, and system for pushing financial products to agricultural enterprises. Background Technology

[0002] Currently, financial products offered by banks and other financial institutions do not employ refined classification management for agricultural enterprises. They merely determine the industry to which the customer belongs, and there is no other difference between them and ordinary customers. The ability to identify whether a user is involved in agriculture is low, resulting in unsatisfactory marketing effects for agricultural financial products. Summary of the Invention

[0003] One objective of this invention is to provide a method for recommending financial products to agricultural enterprises, thereby improving the accuracy of identifying agricultural enterprises and increasing the success rate of financial product transactions. Another objective of this invention is to provide a device for recommending financial products to agricultural enterprises. A further objective of this invention is to provide a system for recommending financial products to agricultural enterprises. A further objective of this invention is to provide a computer device. A still further objective of this invention is to provide a readable medium.

[0004] To achieve the above objectives, one aspect of the present invention discloses a method for promoting financial products to agricultural enterprises, comprising:

[0005] Receive model parameters transmitted from multiple enterprise terminals, wherein the model parameters are the model parameters of the enterprise model obtained by each enterprise terminal training a preset model based on internal enterprise data;

[0006] The joint model parameters of the preset model are obtained based on the model parameters of multiple enterprise terminals, and the agricultural identification model is obtained based on the joint model parameters and the preset model.

[0007] The agricultural identification model is used to predict user information to obtain agricultural identification results. Based on the agricultural identification results and user data, a preset financial discount prediction model is used to obtain the user's discount success rate. Push information corresponding to the discount success rate is then determined and pushed to the user's terminal.

[0008] Preferably, obtaining the joint model parameters of the preset model based on the model parameters of multiple enterprise terminals specifically includes:

[0009] Determine the weights corresponding to the model parameters;

[0010] The joint model parameters are determined based on the parameters of each model and their corresponding weights.

[0011] Preferably, determining the joint model parameters based on each model parameter and its corresponding weight specifically includes:

[0012] The joint model parameters are obtained by multiplying the model parameters of all enterprise models corresponding to each model parameter by their corresponding weights and then summing the results.

[0013] Preferably, obtaining the agricultural identification model based on the joint model parameters and the preset model specifically includes:

[0014] The agricultural identification model is obtained by replacing the corresponding parameters in the preset model with the parameters of the joint model.

[0015] Preferably, it further includes, before receiving model parameters transmitted from multiple enterprise terminals:

[0016] The model feature table is sent to each enterprise terminal. The model feature table includes agricultural features and corresponding feature values ​​so that each enterprise terminal can extract features from local data according to the model feature table to obtain feature vectors, and train a preset model according to the feature vectors to obtain the enterprise model.

[0017] Preferably, it further includes, before receiving model parameters transmitted from multiple enterprise terminals:

[0018] Encrypt the user ID in the user data to obtain the encrypted ID;

[0019] The encrypted ID is transmitted to the enterprise terminal so that the enterprise terminal can determine the corresponding internal enterprise data based on the encrypted ID.

[0020] This invention also discloses a method for pushing financial products to agricultural enterprises, including:

[0021] The enterprise model is trained using internal enterprise data to obtain a preset model. The model parameters of the enterprise model are transmitted to the agricultural enterprise financial product push device so that the agricultural enterprise financial product push device obtains the joint model parameters of the preset model based on the model parameters of multiple enterprise terminals. Based on the joint model parameters and the preset model, an agricultural identification model is obtained. The agricultural identification model is used to predict users to obtain agricultural identification results. Based on the agricultural identification results and user data, the user discount success rate is obtained through a preset financial discount prediction model. Push information corresponding to the discount success rate is determined and pushed to the user terminal.

[0022] This invention also discloses a device for pushing financial products to agricultural enterprises, comprising:

[0023] The information receiving module is used to receive model parameters transmitted by multiple enterprise terminals. The model parameters are the model parameters of the enterprise model obtained by each enterprise terminal training a preset model based on internal enterprise data.

[0024] The joint modeling module is used to obtain joint model parameters of a preset model based on model parameters from multiple enterprise terminals, and to obtain an agricultural identification model based on the joint model parameters and the preset model.

[0025] The product push module is used to predict the user's agricultural identification result through the agricultural identification model, obtain the user's discount success rate through a preset financial discount prediction model based on the agricultural identification result and user data, and determine the push information corresponding to the discount success rate to push to the user terminal.

[0026] This invention also discloses an enterprise terminal configured to train an enterprise model based on internal enterprise data to obtain a preset model, transmit the model parameters of the enterprise model to an agricultural enterprise financial product push device so that the agricultural enterprise financial product push device obtains joint model parameters of the preset model based on the model parameters of multiple enterprise terminals, and obtains an agricultural identification model based on the joint model parameters and the preset model; predicts users using the agricultural identification model to obtain an agricultural identification result, obtains the user discount success rate based on the agricultural identification result and user data through a preset financial discount prediction model, and determines push information corresponding to the discount success rate to push to the user terminal.

[0027] The present invention also discloses a financial product push system for agricultural enterprises, including a financial product push device for agricultural enterprises and a user terminal;

[0028] The agricultural enterprise financial product push device is used to receive model parameters transmitted from multiple enterprise terminals. The model parameters are model parameters of the enterprise model obtained by each enterprise terminal training a preset model based on internal enterprise data. The device obtains joint model parameters of the preset model based on the model parameters of multiple enterprise terminals. The device obtains an agricultural identification model based on the joint model parameters and the preset model. The device predicts the user's agricultural identification result using the agricultural identification model. The device obtains the user's discount success rate based on the agricultural identification result and user data using a preset financial discount prediction model. The device determines the push information corresponding to the discount success rate and pushes it to the user terminal.

[0029] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0030] The present invention also discloses a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0031] This invention discloses a method for pushing financial products to agricultural enterprises. The method receives model parameters transmitted from multiple enterprise terminals. These model parameters are model parameters of an enterprise model obtained by each enterprise terminal training a preset model based on its internal data. It then obtains joint model parameters of the preset model based on the model parameters from multiple enterprise terminals, and finally obtains an agricultural identification model based on the joint model parameters and the preset model. The method uses this agricultural identification model to predict user behavior and obtain an agricultural identification result. Based on the agricultural identification result and user data, it uses a preset financial discount prediction model to obtain the user's discount success rate, and determines the push information corresponding to the discount success rate to push to the user terminal. Thus, this invention obtains an enterprise model by training a preset model using internal data from external enterprises, acquires model parameters transmitted from external enterprise terminals, performs vertical federated learning based on the model parameters from multiple enterprise terminals to obtain an agricultural identification model, uses this agricultural identification model to predict user behavior and obtain an agricultural identification result, predicts the discount success rate based on the agricultural identification result and user data, and pushes financial products based on the discount success rate, thereby improving the success rate of financial product transactions. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 shows a structural diagram of a specific embodiment of the financial product push system for agricultural enterprises of the present invention;

[0034] Figure 2 shows a flowchart of a specific embodiment of the method for pushing financial products to agricultural enterprises according to the present invention;

[0035] Figure 3 shows a flowchart of a specific embodiment S200 of the method for pushing financial products to agricultural enterprises according to the present invention;

[0036] Figure 4 shows a schematic diagram of the construction of the agricultural identification model in a specific embodiment of the method for pushing financial products to agricultural enterprises according to the present invention;

[0037] Figure 5 shows a flowchart of data alignment in a specific embodiment of the method for pushing financial products to agricultural enterprises according to the present invention;

[0038] Figure 6 shows a structural diagram of a specific embodiment of the agricultural enterprise financial product delivery device of the present invention;

[0039] Figure 7 shows a schematic diagram of a computer device suitable for implementing embodiments of the present invention. Detailed Implementation

[0040] 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.

[0041] The terms "first," "second," etc., used in this document do not specifically refer to any order or sequence, nor are they intended to limit the invention; they are merely used to distinguish elements or operations described using the same technical terms.

[0042] The terms “include,” “including,” “have,” “contain,” etc., used in this article are all open-ended terms, meaning that they include but are not limited to.

[0043] The term "and / or" as used herein includes any or all of the things mentioned.

[0044] It should be noted that the acquisition, storage, use, and processing of data in the technical solution of this invention all comply with the relevant provisions of national laws and regulations.

[0045] It should be noted that the method, apparatus and system for pushing financial products to agricultural enterprises disclosed in this application can be used in the field of artificial intelligence technology, or in any field other than artificial intelligence technology. The application field of the method, apparatus and system for pushing financial products to agricultural enterprises disclosed in this application is not limited.

[0046] To facilitate understanding of the technical solution provided in this application, the relevant content of the technical solution is explained below. The method for pushing agricultural financial products to enterprises provided in this application obtains an enterprise model by training a preset model using internal data from external enterprise enterprises. Model parameters transmitted from external enterprise terminals are obtained from the enterprise terminals. Then, vertical federated learning is performed based on the model parameters from multiple enterprise terminals to obtain an agricultural identification model. This agricultural identification model is used to predict user behavior and obtain agricultural identification results. Based on the agricultural identification results and user data, a discount success rate is predicted. Financial products are then pushed based on the discount success rate, thereby improving the success rate of financial product transactions.

[0047] Figure 1 is a schematic diagram of the structure of the agricultural enterprise financial product push system provided in the embodiment of this application. As shown in Figure 1, the agricultural enterprise financial product push system provided in the embodiment of this application includes an agricultural enterprise financial product push device 1 and a user terminal 2.

[0048] The agricultural enterprise financial product push device 1 is used to receive model parameters transmitted from multiple enterprise terminals. The model parameters are the model parameters of the enterprise model obtained by each enterprise terminal training a preset model based on internal enterprise data. The device obtains joint model parameters of the preset model based on the model parameters of multiple enterprise terminals. The device obtains an agricultural identification model based on the joint model parameters and the preset model. The device predicts the user's agricultural identification result through the agricultural identification model. The device obtains the user's discount success rate through a preset financial discount prediction model based on the agricultural identification result and user data. The device determines the push information corresponding to the discount success rate and pushes it to the user terminal 2.

[0049] The following uses the agricultural enterprise financial product push device 1 as an example to illustrate the implementation process of the agricultural enterprise financial product push method provided in this application embodiment. It is understood that the execution entity of the agricultural enterprise financial product push method provided in this application embodiment includes, but is not limited to, the agricultural enterprise financial product push device 1.

[0050] This invention discloses a method for pushing financial products to agricultural enterprises. As shown in Figure 2, in this embodiment, the method includes:

[0051] S100: Receive model parameters transmitted by multiple enterprise terminals, wherein the model parameters are the model parameters of the enterprise model obtained by each enterprise terminal training a preset model based on internal enterprise data.

[0052] S200: Obtain the joint model parameters of the preset model based on the model parameters of multiple enterprise terminals, and obtain the agricultural identification model based on the joint model parameters and the preset model.

[0053] S300: The agricultural identification model is used to predict the user's agricultural identification result. Based on the agricultural identification result and user data, the user's discount success rate is obtained through a preset financial discount prediction model. The push information corresponding to the discount success rate is determined and pushed to the user's terminal.

[0054] This invention discloses a method for pushing financial products to agricultural enterprises. The method receives model parameters transmitted from multiple enterprise terminals. These model parameters are model parameters of an enterprise model obtained by each enterprise terminal training a preset model based on its internal data. It then obtains joint model parameters of the preset model based on the model parameters from multiple enterprise terminals, and finally obtains an agricultural identification model based on the joint model parameters and the preset model. The method uses this agricultural identification model to predict user behavior and obtain an agricultural identification result. Based on the agricultural identification result and user data, it uses a preset financial discount prediction model to obtain the user's discount success rate, and determines the push information corresponding to the discount success rate to push to the user terminal. Thus, this invention obtains an enterprise model by training a preset model using internal data from external enterprises, acquires model parameters transmitted from external enterprise terminals, performs vertical federated learning based on the model parameters from multiple enterprise terminals to obtain an agricultural identification model, uses this agricultural identification model to predict user behavior and obtain an agricultural identification result, predicts the discount success rate based on the agricultural identification result and user data, and pushes financial products based on the discount success rate, thereby improving the success rate of financial product transactions.

[0055] In a preferred embodiment, as shown in Figure 3, step S200, which obtains the joint model parameters of the preset model based on the model parameters of multiple enterprise terminals, specifically includes:

[0056] S210: Determine the weights corresponding to the model parameters.

[0057] S220: Determine the joint model parameters based on the parameters of each model and their corresponding weights.

[0058] Specifically, when constructing an agricultural identification model based on federated learning, it is necessary to combine the model parameters transmitted from each enterprise terminal to obtain the joint model parameters of the agricultural identification model. In this preferred embodiment, by determining the weights of the model parameters, the model parameters can be weighted and then combined to determine the joint model parameters. It should be noted that the weights corresponding to the model parameters can be determined based on historical agricultural transaction data using big data algorithms to determine the importance of each model parameter, thereby determining the weights corresponding to each model parameter. Of course, the model parameters and their corresponding weights can also be determined according to actual needs, and this application does not limit this.

[0059] In a preferred embodiment, S220, which determines the joint model parameters based on each model parameter and its corresponding weight, specifically includes:

[0060] S221: Multiply the model parameters of all enterprise models corresponding to each model parameter by their corresponding weights and then add them together to obtain the joint model parameters.

[0061] In this preferred embodiment, when determining the joint model parameters by weighting the model parameters, the model parameters transmitted by all enterprise terminals corresponding to the same model parameters can be weighted and summed to obtain the final joint model parameters that can be used for agricultural identification models.

[0062] In a preferred embodiment, step S200, which obtains the agricultural identification model based on the joint model parameters and the preset model, specifically includes:

[0063] S230: The agricultural identification model is obtained by replacing the corresponding parameters in the preset model with the joint model parameters.

[0064] Specifically, as shown in Figure 4, when the enterprise terminal includes enterprise A, enterprise B, and enterprise C, each enterprise, after data alignment, trains corresponding enterprise models A, B, and C. Furthermore, joint model parameters can be obtained based on the model parameters transmitted from models A, B, and C. These joint model parameters are then used to replace the corresponding parameters in the preset model to obtain the agricultural identification model.

[0065] In a preferred embodiment, the method further includes, before receiving model parameters transmitted from multiple enterprise terminals:

[0066] S010: Send the model feature table to each enterprise terminal. The model feature table includes agricultural features and corresponding feature values ​​so that each enterprise terminal can extract features from local data according to the model feature table to obtain feature vectors, and train a preset model according to the feature vectors to obtain the enterprise model.

[0067] Specifically, it is understandable that, to ensure the security of internal data within companies outside the bank, this internal data cannot be obtained from external companies. Therefore, if a bank relies solely on its internal user data to build an agricultural identification model to determine whether a user is involved in agriculture, the accuracy of the identification may be low, leading to failed financial product transactions. Based on this, this invention employs vertical federated learning, where each company trains a pre-defined model based on its internal data to obtain a company model. This model training process includes extracting feature vectors from the company's internal data based on pre-defined features, and then inputting these feature vectors and corresponding agricultural-related labels into the pre-defined model to train the company model.

[0068] In this process, in order for the model parameters of the enterprise model to be used to build the bank's agricultural identification model, the bank needs to pre-set features and send the features to the enterprise terminal so that the enterprise terminal can extract features from the enterprise's internal data based on the features.

[0069] In specific examples, external companies may include at least one of the following: telecommunications data, pesticide companies, seed companies, forestry companies, and feed companies.

[0070] The characteristics of user data within banks may include: customer star rating, customer name, group number, group name, industry, administrative division code, company name (English), licensed business items, business scope, enterprise (institution) type code, economic sector code, economic sector name, customer industry logo, enterprise size (large, medium, small, micro), green credit, industry policy classification, advanced manufacturing, address, administrative division of residence, city, district / county, customer business affiliation, supply chain, enterprise size, listed companies, overseas / domestic enterprises, industry policies, import / export enterprises, financing relationships, credit trends, industry risk, cross-industry operations, concentrated transactions with related companies, frequent fund transfers to the same customer, debt repayment ability, financial efficiency, development capability, complex related types, counterparty risk, small business investment in microfinance companies, and legal representative. Frequent changes, frequent changes of business location, small business management identification, blacklist for safe production [enterprise], import and export credit (import and export credit - import and export administrative penalties), list of abnormal operations, product quality spot checks failures, negative list [enterprise], A-level taxpayer [enterprise], list of customs advanced certified enterprises, enterprise environmental credit rating, environmental protection rating, enterprise tax rating - tax rating, import and export credit (import and export credit - import and export credit rating), channel preference, listed company identification, consumer finance identification, auto finance identification, finance company identification, internal caliber small business identification, internal caliber small business identification, regulatory caliber small business identification, industry, enterprise size, economic composition, customer status, current credit rating, small business management identification, customer name, establishment date, enterprise size, customer green credit classification, business scope, customer financing twelve-level classification, enterprise size, small business management identification, number of fund transactions, enterprise online banking transaction amount, credit limit usage, etc., at least one of the following characteristics.

[0071] The characteristics of a pesticide company may include at least one of the following: the number of times pesticides were purchased in one month, the total amount of pesticides purchased in one month, the types of pesticides purchased in one month, the weight of pesticides purchased in one month, the number of times pesticides were purchased in three months, the total amount of pesticides purchased in three months, the types of pesticides purchased in three months, the weight of pesticides purchased in three months, the number of times pesticides were purchased in six months, the total amount of pesticides purchased in six months, the types of pesticides purchased in six months, the weight of pesticides purchased in six months, the number of times pesticides were purchased in twelve months, the total amount of pesticides purchased in twelve months, the types of pesticides purchased in twelve months, the weight of pesticides purchased in twelve months, the number of times pesticides were purchased in 24 months, the total amount of pesticides purchased in 24 months, the types of pesticides purchased in 24 months, and the weight of pesticides purchased in 24 months.

[0072] The fertilizer company possesses at least one of the following characteristics: number of fertilizer purchases within 1 month, total amount of fertilizer purchases within 1 month, type of fertilizer purchased within 1 month, weight of fertilizer purchased within 1 month, number of fertilizer purchases within 3 months, total amount of fertilizer purchases within 3 months, type of fertilizer purchased within 3 months, weight of fertilizer purchased within 3 months, number of fertilizer purchases within 6 months, total amount of fertilizer purchases within 6 months, type of fertilizer purchased within 6 months, weight of fertilizer purchased within 6 months, number of fertilizer purchases within 12 months, total amount of fertilizer purchases within 12 months, type of fertilizer purchased within 12 months, weight of fertilizer purchased within 12 months, number of fertilizer purchases within 24 months, total amount of fertilizer purchases within 24 months, type of fertilizer purchased within 24 months, and weight of fertilizer purchased within 24 months.

[0073] The characteristics of a seed company may include at least one of the following: number of seed purchases in 1 month, total amount of seed purchases in 1 month, total variety of seed purchases in 1 month, weight of seed purchases in 1 month, number of seed purchases in 3 months, total amount of seed purchases in 3 months, total variety of seed purchases in 3 months, weight of seed purchases in 3 months, number of seed purchases in 6 months, total amount of seed purchases in 6 months, total variety of seed purchases in 6 months, weight of seed purchases in 6 months, number of seed purchases in 12 months, total amount of seed purchases in 12 months, total variety of seed purchases in 12 months, weight of seed purchases in 12 months, number of seed purchases in 24 months, total amount of seed purchases in 24 months, total variety of seed purchases in 24 months, and weight of seed purchases in 24 months.

[0074] The characteristics of a feed company may include at least one of the following: number of feed purchases in 1 month, total amount of feed purchases in 1 month, total types of feed purchased in 1 month, weight of feed purchased in 1 month, number of feed purchases in 3 months, total amount of feed purchases in 3 months, total types of feed purchased in 3 months, weight of feed purchased in 3 months, number of feed purchases in 6 months, total amount of feed purchases in 6 months, total types of feed purchased in 6 months, weight of feed purchased in 6 months, number of feed purchases in 12 months, total amount of feed purchases in 12 months, total types of feed purchased in 12 months, weight of feed purchased in 12 months, number of feed purchases in 24 months, total amount of feed purchases in 24 months, total types of feed purchased in 24 months, and weight of feed purchased in 24 months.

[0075] The characteristics of a telecommunications company may include at least one of the following: the region of origin of shareholder phone numbers, the region of origin of employee phone numbers, the region of origin of employee monthly data usage, the region of origin of shareholder frequently used contact phone numbers, the region of origin of employee frequently used contact phone numbers, the region of origin of company office phone numbers, the total duration of company office phone calls (within 1 month), the total duration of company office phone calls (within 3 months), the total duration of company office phone calls (within 6 months), the total duration of company office phone calls (within 12 months), the total duration of company office phone calls (within 24 months), the number of company office phone calls (within 1 month), the number of company office phone calls (within 3 months), the number of company office phone calls (within 6 months), the number of company office phone calls (within 12 months), and the number of company office phone calls (within 24 months).

[0076] Banks can obtain user data from their internal systems. In specific examples, user data may include:

[0077] 1. Historical bill discounting transaction information, which may include bill number, face value, bill type, bill medium, bill maturity date, discount date, discount rate, discounting institution, acceptor, drawer, payee, counterparty, transferability mark, interest payment method, transaction method, business model, etc.

[0078] 2. Historical interest rate quotation information. This includes quotation type, quoting institution, bill type, bill medium, maturity type, maturity value, list of acceptors, list of counterparties, applicable institutions, and floating points.

[0079] 3. Shibor benchmark interest rate information. This includes the type of business, medium of payment, term type, term value, and benchmark interest rate value.

[0080] 4. Bill Disclosure Control Information. This includes the controlling bank number, the institution that set up the control, the control type, the publication date, and the revocation date.

[0081] 5. Public notice and reminder information for negotiable instruments. This includes the instrument number, verification mark, etc.

[0082] 6. Enterprise blacklist information. This includes customer ID (CIS number / Unified Social Credit Code), customer name, control type, control start date, control end date, and control category.

[0083] 7. Sensitive Information. This includes sensitive information types, dictionary codes, dictionary values, etc.

[0084] 8. Bill Circulation Record. This includes the bill number, circulation date, circulation sequence number, transaction type, endorser, and endorsee.

[0085] 9. Authorization Information. This includes the authorizing organization, authorization amount, authorization start date, and authorization expiration date.

[0086] 10. Recovery Information. This includes the type of recovery, the amount recovered, and the party being pursued.

[0087] 11. Other supplementary information, used to link and improve transaction characteristics. This includes information on financial institutions, Shanghai Commercial Paper Exchange (SCPE) institutions, payment system bank names and codes, SCPE institution correspondence information, and other institution information.

[0088] 12. Credit Line Information. This includes customer code, customer name, interbank financing special credit line within one year (inclusive), balance of interbank financing special credit line within one year (inclusive), approved non-special credit line, balance of non-special financing, credit start date, and credit expiration date.

[0089] 13. Industry Competitor List Information. This includes customer ID, list type, list source (whitelist / restricted list), etc.

[0090] 14. Fraudulent information. This includes the type of surveillance system, surveillance system, and surveillance number.

[0091] 15. Customer Information. This includes customer ID, name, customer status, organization code, company size, economic nature, financing investment direction, industry, and location.

[0092] 16. Customer Discount Intention. Includes customer ID, customer name, invoice number, recommendation indicator, probability of discounting within 7 days, probability of discounting within 7 days, probability of discounting within 15 days, probability of discounting within 15 days, probability of discounting within 30 days, probability of discounting within 30 days, probability of discounting within 60 days, probability of discounting within 60 days, etc.

[0093] 17. Credit Information (CIIS). Data tables related to the second-generation credit reporting system.

[0094] 18. Invoice Information (CMAS).

[0095] The characteristics corresponding to user data within a bank can also include basic and derived transaction characteristics. Among these, basic characteristics may include:

[0096] (1) Obtain basic characteristics based on transaction data and mark them as positive or negative samples based on whether the transaction was ultimately successful: bill number, bill type, bill medium, face value, bill maturity date, discount date, discount rate, discounting institution, acceptor number, acceptor name, drawer number, drawer name, payee number, payee name, discount applicant number, discount applicant name, bill transferability identifier, interest payment method, transaction method, business model, and sample mark.

[0097] (2) Based on the combination of transaction data, interest rate quotation data and shibor benchmark interest rate, the following characteristics are obtained: interest rate quotation value, credit flow control value, credit flow release institution level, fund liquidity control value, and fund liquidity release institution level.

[0098] (3) Characteristics are obtained by combining transaction data and public control information.

[0099] (4) Characteristics are obtained by combining transaction data and public notice information.

[0100] (5) Based on transaction data and enterprise blacklist combination, obtain characteristics: whether the drawer is a blacklisted customer, whether the acceptor is a blacklisted customer, whether the payee is a blacklisted customer, and whether the discount applicant is a blacklisted customer.

[0101] (6) Based on the combination of transaction data and sensitive information, the characteristics are obtained: whether the drawer contains sensitive words, whether the acceptor contains sensitive words, whether the payee contains sensitive words, and whether the discount applicant contains sensitive words.

[0102] (7) Based on the combination of transaction data and bill circulation records, the characteristics are obtained: the number of bill endorsements and the consecutive endorsement identifier.

[0103] (8) Based on the combination of transaction data, bill circulation records and enterprise blacklist, obtain the characteristics: whether the historical endorser includes blacklist customers.

[0104] (9) Based on the combination of transaction data, bill circulation records and sensitive information, characteristics are obtained: whether the historical endorser contains sensitive words.

[0105] (10) Based on transaction data and customer discounting intentions, the following characteristics are obtained: discounting probability of the discounting applicant within 7 days, discounting probability of the discounting applicant within 15 days, discounting probability of the discounting applicant within 30 days, and discounting probability of the discounting applicant within 60 days.

[0106] (11) Based on the combination of transaction data and peer list information, the characteristics are obtained: whether the acceptor is on the peer white list or whether the acceptor is on the peer black list.

[0107] (12) Based on the combination of transaction data and credit information, the characteristics are obtained: the non-performing balance of the discount applicant and the balance of the discount applicant under special attention.

[0108] (13) Characteristics obtained from the combination of transaction data and customer information: bank account identifier of drawer, bank account identifier of acceptor, bank account identifier of payee, bank account identifier of discount applicant, enterprise size of drawer, economic nature of drawer, industry of drawer, financing direction of drawer, region of drawer, enterprise size of acceptor, economic nature of acceptor, industry of acceptor, financing direction of acceptor, region of acceptor, enterprise size of payee, economic nature of payee, industry of payee, financing direction of payee, region of payee, enterprise size of discount applicant, economic nature of discount applicant, industry of discount applicant, financing direction of discount applicant, region of discount applicant.

[0109] (14) Characteristics obtained based on a combination of transaction data and credit line information: the acceptor's interbank financing special credit line within one year (inclusive), the acceptor's interbank financing special credit line balance within one year (inclusive), the acceptor's non-special credit line approval limit, the acceptor's non-special financing balance, the acceptor's credit start date, the acceptor's credit expiry date, the discount applicant's non-special credit line approval limit, the discount applicant's non-special financing balance, the discount applicant's credit start date, and the discount applicant's credit expiry date.

[0110] (15) Based on the combination of transaction data and authorization information, the characteristic is obtained: authorized amount.

[0111] (16) Based on the combination of transaction data and recourse information, the characteristics are obtained: the total amount of the acceptor being pursued.

[0112] Derived features may include:

[0113] (1) Based on the characteristics derived from the drawer and the discount applicant: self-opening and self-discounting mark.

[0114] (2) Based on the characteristics derived from the discount rate and the interest rate quote: the interest rate discount points (if the discount rate is greater than the interest rate quote, the discount points are negative).

[0115] (3) Based on the maturity date and discount date of the bill, the characteristic is derived: remaining term.

[0116] Optionally, the default model can be the GBM algorithm. GBM stands for Decision Tree, a tree structure used for classification. Each internal node represents a test decision for a certain attribute, a branch represents a test result, and a leaf represents a class or class distribution. The top-level node is the root node. A decision tree can be understood as a set of if-then rules, with each path from the root node to a leaf node constructing a rule.

[0117] Decision tree learning algorithms include feature selection, decision tree generation, and decision tree pruning. The method for constructing a decision tree is a top-down recursive approach. The construction strategy is as follows: if all samples in the training sample set belong to the same class, they are designated as leaf nodes, and their content represents the class label. Otherwise, an attribute is selected based on a strategy (such as information entropy or the GINI coefficient). The sample set is then divided into subsets according to the attribute's values, ensuring that all samples in each subset have the same attribute value. This process is then repeated recursively for each subset. This approach is essentially the "divide and conquer" principle. The information gain algorithm is used to calculate the information gain value of each leaf node branch, and the feature of the leaf node with the highest information gain is selected as the output.

[0118] Pruning is one of the methods to stop branching in a decision tree. There are two types of pruning: pre-pruning and post-pruning. Pre-pruning involves setting a criterion during the tree's growth process, and stopping growth when that criterion is reached. This approach can easily lead to "vision limitation," meaning that once branching stops and node N becomes a leaf node, any possibility of its successor nodes performing a "good" branching operation is eliminated.

[0119] In post-pruning, the tree first grows sufficiently until all leaf nodes have the minimum impurity value, thus overcoming the "vision limit." Then, for all adjacent pairs of leaf nodes, it is considered whether to eliminate them. If elimination causes a satisfactory increase in impurity, then elimination is performed, and their common parent node becomes the new leaf node. This "merging" of leaf nodes is the opposite of the node branching process. After pruning, leaf nodes are often distributed across a wide level, and the tree becomes unbalanced. The advantage of post-pruning is that it overcomes the "vision limit" effect and does not require retaining some samples for cross-validation, thus fully utilizing the information of the entire training set. However, post-pruning is computationally much more expensive than pre-pruning, especially with large sample sets. Nevertheless, for small sample sets, post-pruning is still superior to pre-pruning.

[0120] Parameter tuning is a process of continuously modeling and searching for better results, followed by further parameter tuning experiments based on the results. In this specific example, the following parameter combination with good experimental results was ultimately selected: learning rate: 0.05, maximum depth of a single tree: 5, number of trees: 100, minimum boost split: 0.0000100000, number of training epochs to stop: 1, stop training baseline: AUTO, stop training tolerance: 0.001.

[0121] In a specific example, when the user discount success rate is obtained by using a preset financial discount prediction model based on the agricultural identification results and user data, and the push information corresponding to the discount success rate is determined and pushed to the user terminal, the interest rate discount level and actual interest rate corresponding to the discount success rate can be determined by referring to the standards in Table 1, thereby forming the corresponding push information and sending it to the user terminal.

[0122] Table 1

[0123] Discount Success Rate, Interest Rate Discount Level, Actual Interest Rate ~0.91%, Interest Rate Discount AAAA Level a ~0.8 0.9%, Interest Rate Discount AAAA Level b ~0.7 0.8%, Interest Rate Discount AAA Level c ~0.6 0.7%, Interest Rate Discount AA Level d ~0.5 0.6%, Interest Rate Discount A Level e surface

[0124] In a preferred embodiment, as shown in Figure 5, the method further includes, before receiving model parameters transmitted from multiple enterprise terminals:

[0125] S021: Encrypt the user ID in the user data to obtain the encrypted ID.

[0126] S022: Transmit the encrypted ID to the enterprise terminal so that the enterprise terminal can determine the corresponding internal enterprise data based on the encrypted ID.

[0127] Specifically, it is understandable that banks and external enterprises can identify shared user data by encrypting user IDs, ensuring data consistency between banks and enterprises. This allows the model parameters of the enterprise models trained by each enterprise terminal to be used to build the bank's internal agricultural identification model.

[0128] In summary, this invention uses federated learning to combine data from both internal and external banks to more accurately identify agricultural enterprises, improve the marketing success rate of bill discounting, expand new marketing channels, and increase customer acquisition capabilities.

[0129] Based on the same principle, this invention discloses a method for pushing financial products to agricultural enterprises. The method includes: training an enterprise model using internal enterprise data to obtain a preset model; transmitting the model parameters of the enterprise model to an agricultural enterprise financial product push device so that the device obtains joint model parameters of the preset model based on model parameters from multiple enterprise terminals; obtaining an agricultural identification model based on the joint model parameters and the preset model; predicting user behavior using the agricultural identification model to obtain an agricultural identification result; obtaining the user's discount success rate using a preset financial discount prediction model based on the agricultural identification result and user data; and determining push information corresponding to the discount success rate and pushing it to the user's terminal.

[0130] Since the principle behind this method is similar to that of the methods described above, the implementation of this method can be found in the implementation of the methods described above, and will not be repeated here.

[0131] Based on the same principle, this invention discloses a device for pushing financial products to agricultural enterprises. As shown in Figure 6, in this embodiment, the device includes an information receiving module 11, a joint modeling module 12, and a product pushing module 13.

[0132] The information receiving module 11 is used to receive model parameters transmitted by multiple enterprise terminals. The model parameters are the model parameters of the enterprise model obtained by each enterprise terminal training a preset model based on internal enterprise data.

[0133] The joint modeling module 12 is used to obtain the joint model parameters of the preset model based on the model parameters of multiple enterprise terminals, and to obtain the agricultural identification model based on the joint model parameters and the preset model.

[0134] The product push module 13 is used to predict the user's agricultural identification result through the agricultural identification model, obtain the user's discount success rate through a preset financial discount prediction model based on the agricultural identification result and user data, and determine the push information corresponding to the discount success rate to push to the user terminal.

[0135] Since the principle by which this device solves the problem is similar to the methods described above, the implementation of this device can refer to the implementation of the methods described above, and will not be repeated here.

[0136] Based on the same principle, this invention discloses an enterprise terminal. The enterprise terminal is used to train an enterprise model using internal enterprise data to obtain a preset model. The model parameters of the enterprise model are transmitted to an agricultural enterprise financial product push device, enabling the agricultural enterprise financial product push device to obtain joint model parameters of the preset model based on model parameters from multiple enterprise terminals. An agricultural identification model is then obtained based on the joint model parameters and the preset model. The agricultural identification model is used to predict user behavior and obtain an agricultural identification result. Based on the agricultural identification result and user data, a preset financial discount prediction model is used to obtain the user's discount success rate. Push information corresponding to the discount success rate is then determined and pushed to the user terminal.

[0137] Since the principle behind this terminal's problem-solving is similar to the methods described above, the implementation of this terminal can refer to the implementation of the methods described above, and will not be repeated here.

[0138] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer device, specifically, a computer device can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0139] In a typical example, a computer device specifically includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method described above.

[0140] Referring now to FIG7, a schematic diagram of the structure of a computer device 700 suitable for implementing embodiments of the present application is shown.

[0141] As shown in Figure 7, the computer device 700 includes a central processing unit (CPU) 701, which can perform various appropriate tasks and processes based on programs stored in read-only memory (ROM) 702 or programs loaded from storage section 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the system 700. The CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0142] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal feedback (LCD), etc., and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed in the storage section 708 as needed.

[0143] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program including program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 709, and / or installed from removable medium 711.

[0144] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0145] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0146] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0147] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0148] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0149] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0150] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0151] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0152] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0153] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A method for promoting financial products to agricultural enterprises, characterized in that, include: The system receives model parameters transmitted from multiple enterprise terminals. These model parameters are the model parameters of an enterprise model obtained by each enterprise terminal training a preset model based on its internal data. The enterprise model is obtained through vertical federated learning, where each enterprise trains the preset model based on its internal data. The enterprise model training process includes extracting feature vectors from the internal data based on preset features, inputting the feature vectors and corresponding "agriculture-related" labels into the preset model to train the enterprise model. The preset model employs the GBM algorithm. Joint model parameters of the preset model are obtained based on the model parameters from multiple enterprise terminals. An agriculture-related identification model is then obtained based on the joint model parameters and the preset model. The agriculture-related identification model is then used for... The model predicts user data to obtain agricultural identification results. Based on the agricultural identification results and user data, a preset financial discount prediction model is used to obtain the user's discount success rate. Push information corresponding to the discount success rate is determined and pushed to the user's terminal. Specifically, obtaining the agricultural identification model based on the joint model parameters and the preset model includes replacing the corresponding parameters in the preset model with the joint model parameters to obtain the agricultural identification model. The method further includes, before receiving model parameters transmitted from multiple enterprise terminals, sending a model feature table to each enterprise terminal. The model feature table includes agricultural features and corresponding feature values ​​so that each enterprise terminal can extract features from its local data based on the model feature table to obtain a feature vector.

2. The method for pushing financial products to agricultural enterprises according to claim 1, characterized in that, The step of obtaining the joint model parameters of the preset model based on the model parameters of multiple enterprise terminals specifically includes: determining the weights corresponding to the model parameters; and determining the joint model parameters based on each model parameter and its corresponding weight.

3. The method for promoting financial products to agricultural enterprises according to claim 2, characterized in that, The specific steps of determining the joint model parameters based on each model parameter and its corresponding weights include: multiplying the model parameters of all enterprise models corresponding to each model parameter by their corresponding weights and then summing them to obtain the joint model parameters.

4. The method for promoting financial products to agricultural enterprises according to claim 1, characterized in that, The method further includes, before receiving model parameters transmitted from multiple enterprise terminals: encrypting the user ID in the user data to obtain an encrypted ID; and transmitting the encrypted ID to the enterprise terminal so that the enterprise terminal can determine the corresponding internal enterprise data based on the encrypted ID.

5. A method for promoting financial products to agricultural enterprises, characterized in that, include: An enterprise model is obtained by training a preset model based on internal enterprise data. The model parameters of the enterprise model are transmitted to an agricultural enterprise financial product push device. This device then obtains joint model parameters of the preset model based on model parameters from multiple enterprise terminals. An agricultural identification model is obtained based on the joint model parameters and the preset model. The agricultural identification model is used to predict user behavior and obtain an agricultural identification result. Based on the agricultural identification result and user data, a preset financial discount prediction model is used to obtain the user's discount success rate. Push information corresponding to the discount success rate is then determined and pushed to the user's terminal. The enterprise model is obtained through vertical federated learning, where each enterprise trains the preset model based on its internal data. The process includes extracting feature vectors from internal enterprise data based on preset features, inputting the feature vectors and corresponding agricultural-related labels into a preset model to train the preset model and obtain an enterprise model; the preset model adopts the GBM algorithm; obtaining the agricultural-related identification model based on the joint model parameters and the preset model specifically includes: replacing the corresponding parameters in the preset model with the joint model parameters to obtain the agricultural-related identification model; wherein, the agricultural enterprise financial product push device is further used to send a model feature table to each enterprise terminal before receiving model parameters transmitted by multiple enterprise terminals, the model feature table including agricultural-related features and corresponding feature values ​​so that each enterprise terminal can extract features from local data based on the model feature table to obtain a feature vector.

6. A device for pushing financial products to agricultural enterprises, characterized in that, include: An information receiving module is used to receive model parameters transmitted from multiple enterprise terminals. These model parameters are the model parameters of an enterprise model obtained by each enterprise terminal training a preset model based on its internal data. The enterprise model is obtained through vertical federated learning, where each enterprise trains the preset model based on its internal data. The enterprise model training process includes extracting feature vectors from the internal data based on preset features, inputting the feature vectors and corresponding "agriculture-related" labels into the preset model to train the enterprise model. The preset model uses the GBM algorithm. A joint modeling module is used to obtain joint model parameters of the preset model based on the model parameters from multiple enterprise terminals, and to obtain an agriculture-related identification model based on the joint model parameters and the preset model. A product push module is used... The system uses the agricultural identification model to predict user information and obtain agricultural identification results. Based on the agricultural identification results and user data, it uses a preset financial discount prediction model to obtain the user's discount success rate and determines the push information corresponding to the discount success rate to push to the user terminal. Specifically, obtaining the agricultural identification model based on the joint model parameters and the preset model includes replacing the corresponding parameters in the preset model with the joint model parameters to obtain the agricultural identification model. The agricultural enterprise financial product push device is further used to send a model feature table to each enterprise terminal before receiving model parameters transmitted by multiple enterprise terminals. The model feature table includes agricultural features and corresponding feature values ​​so that each enterprise terminal can extract features from local data based on the model feature table to obtain a feature vector.

7. An enterprise terminal, characterized in that, The system is configured to train a preset model based on internal enterprise data to obtain an enterprise model. The model parameters of this enterprise model are then transmitted to an agricultural enterprise financial product push device. This device obtains joint model parameters of the preset model based on model parameters from multiple enterprise terminals. An agricultural identification model is then derived based on these joint model parameters and the preset model. The agricultural identification model is used to predict user behavior and obtain an agricultural identification result. Based on the agricultural identification result and user data, a preset financial discount prediction model is used to obtain the user's discount success rate. Push information corresponding to the discount success rate is then determined and pushed to the user's terminal. The enterprise model is obtained through vertical federated learning, where each enterprise trains the preset model based on its internal data. The training process includes extracting feature vectors from internal enterprise data based on preset features, inputting the feature vectors and corresponding agricultural-related labels into a preset model to train the preset model and obtain an enterprise model; the preset model adopts the GBM algorithm; obtaining the agricultural-related identification model based on the joint model parameters and the preset model specifically includes: replacing the corresponding parameters in the preset model with the joint model parameters to obtain the agricultural-related identification model; the agricultural enterprise financial product push device is further used to send a model feature table to each enterprise terminal before receiving model parameters transmitted by multiple enterprise terminals, the model feature table including agricultural-related features and corresponding feature values ​​so that each enterprise terminal can extract features from local data based on the model feature table to obtain a feature vector.

8. A financial product delivery system for agricultural enterprises, characterized in that, The system includes a device for pushing financial products to agricultural enterprises and a user terminal. The device receives model parameters transmitted from multiple enterprise terminals. These model parameters are the model parameters of an enterprise model obtained by each enterprise terminal training a preset model based on its internal data. It then obtains joint model parameters of the preset model based on the model parameters from the multiple enterprise terminals. Based on the joint model parameters and the preset model, it obtains an agricultural identification model. The agricultural identification model is used to predict user behavior and obtain an agricultural identification result. Based on the agricultural identification result and user data, a preset financial discount prediction model is used to obtain the user's discount success rate. Push information corresponding to the discount success rate is then determined and pushed to the user terminal. The enterprise model is obtained through vertical federated learning, where each enterprise trains the preset model based on its internal data. The training process of the enterprise model includes extracting features from internal enterprise data based on preset features to obtain feature vectors, inputting the feature vectors and corresponding whether or not they are related to agriculture into a preset model to train the preset model and obtain the enterprise model; the preset model adopts the GBM algorithm; the step of obtaining the agricultural identification model based on the joint model parameters and the preset model specifically includes: replacing the corresponding parameters in the preset model with the joint model parameters to obtain the agricultural identification model; the agricultural enterprise financial product push device is further used to send a model feature table to each enterprise terminal before receiving the model parameters transmitted by multiple enterprise terminals, the model feature table including agricultural features and corresponding feature values ​​so that each enterprise terminal can extract features from local data based on the model feature table to obtain feature vectors.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-5.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-5.

Images