Post-loan fund monitoring method, device and equipment, medium and program product

Abstract

The invention discloses a post-loan fund monitoring method and device, equipment, a medium and a program product, and relates to a block chain technology. The method comprises the following steps: acquiring consumption feature information of a target loan user through a block chain; determining consumption characteristic index parameters according to the consumption characteristic information and a pre-trained characteristic analysis model; and according to the consumption characteristic index parameters and a pre-trained risk assessment model, determining post-loan risk factors of the target loan user, and monitoring post-loan funds of the target loan user based on the post-loan risk factors. According to the method and the device, the abnormal flow direction of the loan fund can be identified in time based on the block chain technology, so that bad loan and irregular operation can be prevented, and the safety and compliance of the loan fund are ensured.

Description

Technical Field

[0001] The present invention relates to the field of blockchain technology, and in particular to a post-loan fund monitoring method, device, equipment, medium and program product. Background Art

[0002] With the development of the economy, people's income levels have generally increased, and the demand for improving the quality of life has also increased. Consumer loans enable people to enjoy goods and services in advance without having full immediate payment capabilities. Consumer loans are a form of loan specifically used for daily consumer expenditures of individuals or families. It allows borrowers to obtain a sum of money to meet specific consumer needs, but cannot be used for high-risk investment areas.

[0003] Post-loan fund monitoring refers to the process of tracking and supervising the use of funds by borrowers after the loan is issued. This process aims to ensure that the borrower uses the loan in accordance with the purpose agreed in the contract and can repay the loan principal and interest on time. Therefore, banks not only need to conduct detailed credit assessment and risk control on borrowers before issuing loans, but also need to closely monitor the flow of loan funds after the loan is issued to prevent consumer loan funds from being diverted to high-risk investment areas, such as stocks, real estate and futures, to ensure the safety and compliance of loan funds. However, the flow of post-loan funds often involves multiple departments, and it is difficult to fully track the flow of funds based on the bank's own data alone. Summary of the invention

[0004] The present invention provides a post-loan fund monitoring method, device, equipment, medium and program product, which can timely identify the abnormal flow of loan funds based on blockchain technology, thereby preventing the occurrence of bad loans and non-compliant operations and ensuring the security and compliance of loan funds.

[0005] In a first aspect, an embodiment of the present application provides a method for monitoring post-loan funds, the method comprising:

[0006] Obtain consumption characteristic information of target loan users through blockchain;

[0007] Determine consumption characteristic index parameters according to the consumption characteristic information and a pre-trained characteristic analysis model; wherein the characteristic analysis model is trained using a gradient boosting decision tree as the original model and using consumption characteristic information of historical loan users as input data;

[0008] According to the consumption characteristic indicator parameters and a pre-trained risk assessment model, the post-loan risk factor of the target loan user is determined, and the post-loan funds of the target loan user are monitored based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and the consumption characteristic indicator parameters of historical loan users as input data.

[0009] In a second aspect, the embodiment of the present application provides a post-loan fund monitoring device, the device comprising: an information acquisition module, a parameter determination module and a risk monitoring module; wherein:

[0010] The information acquisition module is used to obtain consumption characteristic information of the target loan user through the blockchain;

[0011] The parameter determination module is used to determine the consumption characteristic index parameters according to the consumption characteristic information and a pre-trained characteristic analysis model; wherein the characteristic analysis model is trained by using a gradient boosting decision tree as the original model and using the consumption characteristic information of historical loan users as input data;

[0012] The risk monitoring module is used to determine the post-loan risk factor of the target loan user based on the consumption characteristic indicator parameters and a pre-trained risk assessment model, and monitor the post-loan funds of the target loan user based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and consumption characteristic indicator parameters of historical loan users as input data.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0014] one or more processors;

[0015] a memory for storing one or more programs,

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the post-loan funds monitoring method described in any embodiment of the present application.

[0017] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, and when the program is executed by a processor, the post-loan fund monitoring method described in any embodiment of the present application is implemented.

[0018] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the post-loan funds monitoring method described in any embodiment of the present application.

[0019] The monitoring method of post-loan funds provided by the embodiment of the present invention obtains the consumption characteristic information of the target loan user; determines the consumption characteristic index parameters according to the consumption characteristic information and the pre-trained characteristic analysis model; determines the post-loan risk factor of the target loan user according to the consumption characteristic index parameters and the pre-trained risk assessment model. In the above technical scheme, on the one hand, the data originally scattered in various institutions are applied with blockchain technology, and are completely obtained through compliant means, so that the source and flow of the data are clearer, which provides important data support for the subsequent evaluation of the use of post-loan funds by loan users. On the other hand, the consumption characteristic information of the target loan user obtained from the blockchain is screened, analyzed and combined using the characteristic analysis model, so that the characteristic processing and accurate analysis of the consumption characteristic information are realized, and the consumption characteristic index parameters with strong explanatory power for the consumption characteristic information are generated. This application solves the problem that the existing technology cannot effectively supervise the flow of post-loan funds, and can timely identify the abnormal flow of loan funds based on blockchain technology, so as to prevent the occurrence of bad loans and non-compliant operations, and ensure the security and compliance of loan funds.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flowchart of a method for monitoring post-loan funds provided in one embodiment of the present application;

[0022] Figure 2 A schematic flow chart of a method for monitoring post-loan funds provided by another embodiment of the present invention;

[0023] Figure 3 A flowchart of a method for training a feature analysis model provided in one embodiment of the present application;

[0024] Figure 4 A flowchart of a risk assessment model training method provided in an example of the present application;

[0025] Figure 5 A schematic diagram of the structure of a monitoring device for post-loan funds provided in one embodiment of the present application;

[0026] Figure 6 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0027] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only the parts related to the present application, rather than all structures, are shown in the accompanying drawings.

[0028] Figure 1 This is a flow chart of a method for monitoring post-loan funds provided in an embodiment of the present application. The method can be performed by a post-loan funds monitoring device or electronic device. The device or electronic device can be implemented in software and / or hardware. The device or electronic device can be integrated in any intelligent device with network communication function. Figure 1 As shown, the post-loan fund monitoring method may include the following steps:

[0029] S101. Obtain consumption characteristic information of target loan users through blockchain.

[0030] The consumption characteristic information is used to indicate characteristic information related to consumption activities of the target loan user during the consumption period. In this embodiment, the consumption characteristic information includes various fund flow data. For example, the inflow and outflow of funds, the proportion of funds inflow and outflow, the channels of inflow and outflow of funds, the residence time and flow information of funds, and the basic information of the target loan user.

[0031] It is worth noting that the information collected in this embodiment is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with the relevant laws, regulations and standards of relevant countries and regions, take necessary confidentiality measures, do not violate public order and good customs, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0032] Specifically, since banks currently use blockchain technology to establish blockchain alliances with other financial institutions, regulators and third-party service providers, they can prevent people's abnormal consumption in a timely manner. Therefore, the consumption characteristics of target loan users can be obtained from blockchain technology through compliant means. The data obtained from the blockchain can be constructed as a transaction network graph. By performing different graph data processing on the transaction network graph, the consumption characteristics of the target loan users can be obtained.

[0033] Among them, the methods of constructing transaction network graphs using blockchain include network topology, node discovery mechanism or graph modeling, etc. Graph data processing of transaction network graphs can include node embedding algorithms, structured feature construction and data cleaning, etc.

[0034] In this embodiment, blockchain technology is applied to the data originally scattered among various institutions, and the data is completely obtained through compliant means. The source and flow of the data are clearer, providing important data support for the subsequent evaluation of the loan users' use of post-loan funds.

[0035] S102. Determine consumption characteristic indicator parameters based on consumption characteristic information and a pre-trained characteristic analysis model.

[0036] The feature analysis model is trained by using the gradient boosting decision tree model as the original model and the consumption feature information of historical loan users as input data. The consumption feature index parameter may include vector feature information corresponding to the consumption feature information and the consumption feature importance index.

[0037] Specifically, after obtaining the consumption feature information of the target loan user, the consumption feature information can be input into the model as the input data of the feature analysis model. The original model of the feature analysis model is the Gradient Boosting Decision Tree (GBDT) model, which can be regarded as a "feature processor" that analyzes the original features of the input data and performs feature division, feature combination and feature selection to generate feature indicators and vector features that are more important to the input data. Therefore, the consumption feature information is input into the feature analysis model, and the feature analysis model can analyze the consumption feature information and output consumption feature indicator parameters. Among them, the consumption feature importance index can indicate which specific consumption feature of the target loan user has a greater risk of post-loan fund use. The vector feature information is a new feature information combination obtained after analyzing, selecting and combining each specific consumption feature of the target loan user. The feature information combination is presented in the form of a vector.

[0038] In this embodiment, the consumption characteristic information of the target loan user obtained is analyzed by using a pre-trained characteristic analysis model, and the consumption characteristic index parameters are determined, so as to realize the characteristic processing and accurate analysis of the consumption characteristic information, and generate consumption characteristic index parameters with strong explanatory power for the consumption characteristic information. At the same time, the consumption characteristic information of the target loan user obtained from the blockchain is screened, analyzed and combined by using the characteristic analysis model, which solves the problem that the flow of post-loan funds cannot be effectively supervised in the prior art, and realizes the timely screening of important characteristic indicators from the consumption characteristic information of the target loan user, providing more accurate, targeted and efficient data support for the subsequent evaluation of the use of post-loan funds by the loan user.

[0039] S103. Determine the post-loan risk factor of the target loan user based on the consumption characteristic indicator parameters and the pre-trained risk assessment model, and monitor the post-loan funds of the target loan user based on the post-loan risk factor.

[0040] The risk assessment model is trained by using the logistic regression model as the original model and the consumption characteristic index parameters of historical loan users as input data. The post-loan risk factor can be used to indicate the risk level of the target loan user using the post-loan funds. Optionally, the post-loan risk factor can be presented in the form of a post-loan risk probability.

[0041] Specifically, after the consumption characteristic information of the target loan user is preliminarily processed using the characteristic analysis model and the consumption characteristic index parameters are obtained, the consumption characteristic index parameters can be input into the model as the input data of the risk assessment model. The original model of the risk assessment model is the Logistic Regression (LR) model, which can be used as a "scoring model" and is suitable for processing large-scale data, and has a wide range of applications and high computational efficiency. Therefore, by inputting the consumption characteristic index parameters into the pre-trained risk assessment model, the post-loan risk factor of the target loan user can be determined. The post-loan risk factor is used to indicate the risk level of the target loan user using the post-loan funds, such as high-risk investment behavior. In addition, since the more accurate and interpretable consumption characteristic index parameters output by the characteristic analysis model are used before inputting into the risk assessment model, the risk assessment model can more efficiently use the consumption characteristic index parameters to classify and regress these features, thereby significantly improving the performance of the risk assessment model and the accuracy of prediction.

[0042] In this embodiment, the characteristic indicator parameters output by the characteristic analysis model are used as the input parameters of the risk assessment model. The consumption characteristic information of the loan user with strong explanatory power can be used to more accurately determine the post-loan risk factor of the loan user, which solves the problem that the consumption characteristic information obtained from the blockchain technology cannot be quickly analyzed in time to obtain the post-loan fund use risk of the loan user. The characteristic analysis model and the risk assessment model are used to extract key consumption characteristic information from the more complex consumption characteristic information to form characteristic indicator parameters, and the post-loan risk factor of the loan user is determined based on the characteristic indicator parameters, thereby achieving accurate and rapid prediction of the post-loan fund use risk of the loan user, thereby preventing the occurrence of bad loans and non-compliant operations and ensuring the security and compliance of loan funds.

[0043] The monitoring method of post-loan funds provided in the embodiment of the present application obtains the consumption characteristic information of the target loan user; determines the consumption characteristic index parameters according to the consumption characteristic information and the pre-trained characteristic analysis model; determines the post-loan risk factor of the target loan user according to the consumption characteristic index parameters and the pre-trained risk assessment model. In the above technical scheme, on the one hand, the data originally scattered in various institutions are applied with blockchain technology, and are completely obtained through compliant means, so that the source and flow of the data are clearer, which provides important data support for the subsequent evaluation of the post-loan fund use risk of the loan user. On the other hand, the consumption characteristic information of the target loan user obtained from the blockchain is screened, analyzed and combined using the characteristic analysis model, so that the characteristic processing and accurate analysis of the consumption characteristic information are realized, and the consumption characteristic index parameters with strong explanatory power for the consumption characteristic information are generated. At the same time, the problem that the existing technology cannot effectively supervise the flow of post-loan funds is solved, and the important characteristic indicators are timely screened out from the consumption characteristic information of the target loan user, which provides more accurate, targeted and efficient data support for the subsequent evaluation of the post-loan fund use risk of the loan user. Finally, the characteristic indicator parameters output by the characteristic analysis model are used as the input parameters of the risk assessment model. The consumption characteristic information of loan users with strong explanatory power can be used to more accurately determine the post-loan risk factors of loan users, solving the problem that the existing technology cannot effectively supervise the flow of post-loan funds. The characteristic analysis model and the risk assessment model are used to extract key consumption characteristic information from the more complex consumption characteristic information to form characteristic indicator parameters, and the post-loan risk factors of loan users are determined based on the characteristic indicator parameters, achieving accurate and rapid prediction of the post-loan fund use risks of loan users, thereby preventing the occurrence of bad loans and non-compliant operations and ensuring the security and compliance of loan funds.

[0044] Figure 2 This is a flowchart of a method for monitoring post-loan funds provided in another embodiment of the present invention. Based on the above embodiment, this embodiment refines the step of "obtaining consumption characteristic information of target loan users." It is worth noting that the information collected in this embodiment is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with the relevant laws, regulations and standards of relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation portals for users to choose to authorize or refuse. Figure 2 As shown, the monitoring method of post-loan funds may include the following steps:

[0045] S201. Obtain original consumption information of the target loan user.

[0046] Among them, the original consumption information includes basic consumption information and structured information. The original consumption information is the consumption information of the target loan user obtained directly and legally using blockchain technology.

[0047] Specifically, we can obtain information such as the institutions on each block alliance in the blockchain and their corresponding capital flows, that is, obtain information on network nodes and edges in the blockchain, and process the data through data cleaning. After that, we use the graph modeling method to build a graph model. At this point, we can use the constructed graph model as structured information in the original consumption information. In addition, we can use the personal basic information of the target loan user in the blockchain or the bank management institution itself as the basic consumption information in the original consumption information.

[0048] S202. Determine first consumption characteristic information and second consumption characteristic information according to the structured information.

[0049] The structured information in the embodiment of the present application may include at least one type of sub-structure information.

[0050] Specifically, according to the structured information in the original consumption information, the first consumption characteristic information and the second consumption characteristic information can be obtained respectively according to different data processing methods.

[0051] Exemplarily, the first consumption characteristic information can process the structured information using a node embedding algorithm to convert the structured data into low-dimensional vector data. The low-dimensional vector data contains the structural information and relationship information of the graph.

[0052] Exemplarily, the second consumption characteristic information may be determined according to the following steps:

[0053] (1) According to the information type of each sub-structure information and sub-structure information, determine the information type field corresponding to each sub-structure information.

[0054] Specifically, since the structured information includes at least one sub-structure information, for each sub-structure information, the information type field corresponding to each sub-structure information can be determined.

[0055] Exemplarily, the substructure information includes the proportion of funds in and out of each consumption channel, the time the funds stay, and the direction of funds flow. The information type field includes fields related to funds in each channel, fields for the direction of funds flow, fields for the time the funds stay, and fields for the type of funds flow. Among them, the fields related to funds in each channel are determined according to the inflow and outflow of funds and the proportion of inflow and outflow funds to the total funds. The fund flow type field includes various types of fund flows.

[0056] (2) The information type fields corresponding to all sub-structure information are determined as the second consumption characteristic information.

[0057] Specifically, the information type field determined above is used as the second consumption characteristic information.

[0058] S203. Determine third consumption characteristic information based on the basic consumption information.

[0059] Specifically, the basic consumption information may be directly subjected to structured feature extraction, and the information after the structured feature extraction is used as the third consumption feature information.

[0060] S204: Determine the first consumption characteristic information, the second consumption characteristic information, and the third consumption characteristic information as consumption characteristic information.

[0061] Specifically, after obtaining the original consumption information using blockchain technology, the first consumption characteristic information, the second consumption characteristic information and the third consumption characteristic information are determined using different information processing methods, and then the above three types of consumption characteristic information can be determined as consumption characteristic information.

[0062] In this embodiment, the original consumption information is obtained from the blockchain, and different consumption characteristic information is obtained using different data processing methods, so that a large amount of diversified data can be obtained by using different data processing methods, which provides a large amount of input data support for training models or using models later. At the same time, the consumption characteristics of target loan users can be analyzed from different angles or different levels by using different data with the same source data but processed differently, which provides diversified data support for comprehensively determining the post-loan fund use risks of target loan users later.

[0063] S205. Determine consumption characteristic indicator parameters based on consumption characteristic information and a pre-trained characteristic analysis model.

[0064] Figure 3 A flow chart of a method for training a feature analysis model provided in one embodiment of the present application. Figure 3 As shown, the training method of the feature analysis model may include the following steps:

[0065] S301. Obtain a first training set.

[0066] The first training set is a training set for training a feature analysis model. The first training set includes consumption feature information of at least one historical loan user, and the consumption feature information of one historical loan user corresponds to a first label information, and the first label information is used to indicate the importance of the consumption feature information of the historical loan user.

[0067] Specifically, the consumption characteristic information of multiple historical loan users can be obtained by using the above steps S201-S204, and the importance of each consumption characteristic information can be determined based on the consumption characteristic information of each historical loan user, and the importance is used as the first label information.

[0068] S302: Input the consumption characteristic information of the current historical loan user and the first label information corresponding to the consumption characteristic information of the current historical loan user into a gradient boosting decision tree model to obtain the first actual label information of the current historical loan user.

[0069] The first actual label information is used to indicate the importance of the consumption characteristic information of the historical loan users output by the gradient boosting decision tree model.

[0070] Specifically, after obtaining the first training set, the original model, that is, the gradient boosting decision tree model, can be trained using the first training set.

[0071] Exemplarily, the first historical loan user is selected from the first training set as the current historical loan user, and the consumption feature information of the current historical loan user and the first label information corresponding to it are used as input samples and input into the gradient boosting decision tree model. At this time, the gradient boosting decision tree model can output a 1* multi-dimensional sparse vector. The vector is a new feature output by the gradient boosting decision tree model, presented in the form of a vector. In addition, the feature importance index of the gradient boosting decision tree, i.e., the first actual label information, can be output to determine which feature of the current user has a greater impact on the risk of post-loan fund use.

[0072] S303, determining whether the error between the first actual label information and the first label information of the current historical loan user is less than or equal to a first preset threshold; if so, executing S304; if not, executing S305.

[0073] Specifically, it is determined whether the error between the first actual label information and the first label information of the current historical loan user is less than or equal to the first preset threshold. If the error between the first actual label information and the first label information is less than or equal to the first preset threshold, it means that the current accuracy of the model meets the user's needs. Therefore, S304 can be executed. If the error between the first actual label information and the first label information is greater than the first preset threshold, it means that the current accuracy of the model does not meet the user's needs. Therefore, it is necessary to execute S305.

[0074] S304: Determine the gradient boosting decision tree model as the feature analysis model.

[0075] Specifically, if the error between the first actual label information and the first label information is less than or equal to the first preset threshold, it means that the current accuracy of the model meets the user's needs. Therefore, the gradient boosting decision tree model can be determined as the feature analysis model. At this point, the training of the feature analysis model is completed.

[0076] S305 , adjusting the model parameters of the gradient boosting decision tree model, taking the next historical loan user as the current historical loan user, and returning to execute 302 .

[0077] Specifically, if the error between the first actual label information and the first label information is greater than the first preset threshold, it means that the current accuracy of the model does not meet the user's needs. Therefore, the model parameters of the gradient boosting decision tree model can be adjusted, and the next historical loan user is used as the current historical loan user, and the process returns to step S302 for further training.

[0078] Exemplarily, the model parameters of the gradient boosting decision tree model can be adjusted according to the following steps:

[0079] (1) Determine the current parameter combination data of the gradient boosting decision tree model;

[0080] Among them, the current parameter combination data includes: learning rate, number of iterations, maximum depth, minimum sample segmentation, minimum number of leaf node samples and maximum number of features, etc.

[0081] Specifically, since the parameter combination data of the gradient boosting decision tree model can affect the performance of the model, the current parameter combination data of the gradient boosting decision tree model can be determined first.

[0082] (2) Use hyperparameter optimization to optimize the current parameter combination data to obtain the optimized current parameter combination data.

[0083] Among them, the method of hyperparameter optimization can be grid search or random search.

[0084] Specifically, by using a hyperparameter optimization method such as grid search, the optimal parameter combination of the parameter combination data of the gradient boosting decision tree model can be determined at the beginning. In the subsequent training process, if the error between the first actual label information and the first label information is greater than the first preset threshold, the hyperparameter optimization method can be used again to further optimize the current parameter combination data, obtain the optimized current parameter combination data, and conduct the next round of model training.

[0085] In this embodiment, the current parameter combination data is optimized through hyperparameter optimization, so that the parameter optimization process of the model is more accurate and automated.

[0086] Specifically, after the feature analysis model is trained, the consumption feature indicator parameters can be determined based on the consumption feature information and the pre-trained feature analysis model.

[0087] S206. Determine the post-loan risk factor of the target loan user based on the consumption characteristic indicator parameters and the pre-trained risk assessment model, and monitor the post-loan funds of the target loan user based on the post-loan risk factor.

[0088] Figure 4 The following is a flow chart of a training method for a risk assessment model provided in an example of the present application. Figure 4 As shown, the training method of the risk assessment model may include the following steps:

[0089] S401: Obtain a second training set.

[0090] Among them, the second training set includes a consumption characteristic indicator parameter of at least one historical loan user output by the characteristic analysis model, and a consumption characteristic indicator parameter of a historical loan user corresponds to a second label information, and the second label information is used to indicate the post-loan risk indicator of the historical loan user.

[0091] Specifically, when training the risk assessment model, the post-loan risk indicators of historical loan users output by the trained feature analysis model will be obtained as the second training set. For example, the consumption feature information of multiple historical loan users is input into the trained feature analysis model, and a high-dimensional sparse vector array will be output. The high-dimensional sparse vector array corresponds to the consumption feature indicator parameters of multiple historical loan users. In addition, the feature importance indicators of multiple historical loan users will also be output. Therefore, the consumption feature indicator parameters of multiple historical loan users and the feature importance indicators of multiple historical loan users can be used as the second training set for training the risk assessment model, with each historical loan user as the benchmark. Among them, the second label information can be the feature importance indicator output by the feature analysis model. The consumption feature indicator parameter can be a sparse vector corresponding to a historical loan user output by the feature analysis model.

[0092] S402: Input the consumption characteristic indicator parameters of the current historical loan user and the second label information corresponding to the consumption characteristic indicator parameters of the current historical loan user into a logistic regression model to obtain the second actual label information of the current historical loan user.

[0093] The second actual label information is used to indicate the post-loan risk indicator of the historical loan users output by the logistic regression model.

[0094] Specifically, a certain historical loan user in the second training set is used as the current historical loan user. The logistic regression model is used as the original model of the risk assessment model. The consumption characteristic indicator parameters of the current historical loan user and the second label information are input into the logistic regression model together, and the logistic regression model outputs the second actual label information of the current historical loan user.

[0095] S403, determining that the error between the second actual label information of the current historical loan user and the second label information is less than or equal to a second preset threshold; if so, executing S404; if not, executing S405.

[0096] Specifically, it is determined whether the error between the second actual label information and the second label information of the current historical loan user is less than or equal to the second preset threshold. If the error between the second actual label information and the second label information is less than or equal to the second preset threshold, it means that the current accuracy of the model meets the user's needs. Therefore, S404 can be executed. If the error between the second actual label information and the second label information is greater than the second preset threshold, it means that the current accuracy of the model does not meet the user's needs. Therefore, it is necessary to execute S405.

[0097] S404. Determine the logistic regression model as the risk assessment model.

[0098] Specifically, if the error between the second actual label information and the second label information is less than or equal to the second preset threshold, it means that the current accuracy of the model meets the user's needs. Therefore, the logistic regression model can be determined as the risk assessment model. At this point, the training of the risk assessment model is completed.

[0099] S405. Adjust the model parameters of the logistic regression model, take the next historical loan user as the current historical loan user, and return to execute S402.

[0100] Specifically, if the error between the second actual label information and the second label information is greater than the second preset threshold, it means that the current accuracy of the model does not meet the user's needs. Therefore, the model parameters of the logistic regression model can be adjusted, and the next historical loan user is used as the current historical loan user, and the process returns to step S402 for further training.

[0101] Exemplarily, the model parameters of the logistic regression model can be adjusted according to the following steps:

[0102] (1) Determine the current parameter data of the logistic regression model.

[0103] Among them, the current parameter data includes weight, intercept, learning rate, number of iterations, regularization parameter, regularization type, maximum step size, etc.

[0104] Specifically, when the logistic regression model is trained for the first time, the current parameter data will be determined by the user.

[0105] And, each time the model parameters of the logistic regression model need to be adjusted thereafter, the current parameter data of the logistic regression model will be determined.

[0106] (2) Regularize the current parameter data to obtain the regularized current parameter data.

[0107] Specifically, since the regularization parameter and regularization type in the current parameter data of the logistic regression model will affect the fitting of the model, the current parameter data can be regularized, for example, using L1 regularization or L2 regularization in the regularization technology to adjust the current parameter data to obtain the regularized current parameter data, thereby reducing the risk of overfitting.

[0108] In this embodiment, the parameters output from the trained feature analysis model are used as sample data to train the risk assessment model, achieving a combination of high accuracy and strong interpretability of the model. Combining the feature processing capabilities of the gradient boosting decision tree model and the efficient classification performance of the logistic regression model, data with diverse sources and rich feature dimensions are better processed, thus providing a basis for determining accurate post-loan risk factors.

[0109] Optionally, after determining the post-loan risk factor of the target loan user according to the consumption characteristic indicator parameters and the pre-trained risk assessment model, the method further includes:

[0110] (i) Determine the risk threshold and determine whether the post-loan risk factor is greater than or equal to the risk threshold; if so, proceed to (ii); if not, proceed to (iii).

[0111] Among them, the risk threshold is a threshold specified by the user, which is used to determine whether the target loan user's use of loan funds exceeds the controllable range.

[0112] Specifically, after obtaining the post-loan risk factor of the target loan user, the post-loan risk factor will be compared with the risk threshold. If it is determined that the post-loan risk factor is greater than or equal to the risk threshold, it means that the consumption of the target loan user may have uncontrollable risks. Therefore, a risk warning can be sent to the relevant regulatory authorities, and the feature importance index output by the feature analysis model can be sent to the relevant regulatory authorities, so that the relevant regulatory authorities can take timely precautions against non-performing loans and non-compliant operations to ensure the security and compliance of loan funds. If it is determined that the post-loan risk factor is less than the risk threshold, it means that there is no uncontrollable risk in the use of loan funds by the target loan user. Therefore, the relevant regulatory authorities do not need to conduct risk investigations and protections on the target loan users.

[0113] (ii) Send risk warnings to relevant regulatory authorities, and send feature importance indicators output by the feature analysis model to relevant regulatory authorities.

[0114] (iii) Return to execute S201.

[0115] Specifically, after determining that the post-loan risk factor of the target loan user is less than the risk threshold, the target loan user may be replaced, and the process returns to S201 to continue supervising other users.

[0116] It is worth noting that in this embodiment, the target loan user can be one user or multiple users. At the same time, the post-loan fund supervision for the target loan user can be carried out according to a preset cycle. And as the number of loan users increases, the above model can be further optimized regularly.

[0117] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data comply with relevant laws, regulations and standards in relevant regions.

[0118] The monitoring method of post-loan funds provided by the embodiment of the present invention, on the one hand, obtains original consumption information from the blockchain, and uses different data processing methods to obtain different consumption feature information, realizes the use of different data processing methods, obtains a large amount of diversified data, and provides a large amount of input data support for training models or using models later. At the same time, it can also use the same source data but different processed data to analyze the consumption characteristics of target loan users from different angles or different levels, and provide diversified data support for comprehensively determining the use of post-loan funds by target loan users later. In addition, the data obtained from the blockchain can more comprehensively know the consumption characteristics of loan users, while ensuring the security and accuracy of the data. Through the decentralized and tamper-proof characteristics of the blockchain, the integrity and authenticity of the data are ensured, and the effectiveness of the consumption characteristics of loan users is improved. On the other hand, the use of feature analysis models can be responsible for feature engineering and the processing of complex feature relationships, and the use of risk assessment models is responsible for efficient classification and result interpretation. The combination of the two training and use can enable the entire system to better balance performance and interpretability in practical applications, achieve strong generalization ability and interpretability, and facilitate feature screening. Finally, the post-loan risk factor is output through an automated model, thereby reducing the regulatory pressure of bank regulatory authorities. It can not only locate the post-loan fund use risks of target loan users more accurately, but also intercept abnormal use of loan funds in a timely manner to ensure the security and compliance of loan funds.

[0119] Figure 5 This is a schematic diagram of the structure of a monitoring device for post-loan funds provided in one embodiment of the present application. Figure 5 As shown, the post-loan fund monitoring device includes: an information acquisition module 501, a parameter determination module 502 and a risk monitoring module 503; wherein,

[0120] The information acquisition module 501 is used to obtain consumption characteristic information of the target loan user through the blockchain;

[0121] The parameter determination module 502 is used to determine the consumption characteristic index parameters according to the consumption characteristic information and the pre-trained characteristic analysis model; wherein the characteristic analysis model is trained by using the gradient boosting decision tree as the original model and the consumption characteristic information of the historical loan users as the input data;

[0122] The risk monitoring module 503 is used to determine the post-loan risk factor of the target loan user based on the consumption characteristic indicator parameters and a pre-trained risk assessment model, and monitor the post-loan funds of the target loan user based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and consumption characteristic indicator parameters of historical loan users as input data.

[0123] The above-mentioned monitoring device for post-loan funds can execute the method provided by any embodiment of the present application, and has the corresponding functional modules and beneficial effects of the execution method. For technical details not described in detail in this embodiment, please refer to the monitoring method for post-loan funds provided by any embodiment of the present application.

[0124] An embodiment of the present invention also provides a computer program product.

[0125] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer program products, which can include one or more computer programs, which can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor, which can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0126] Figure 6 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, referring to Figure 6 , Figure 6 The electronic device 12 shown is only an example and should not limit the functions and scope of use of the embodiments of the present application. Figure 6 As shown, the electronic device 12 is in the form of a general purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).

[0127] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0128] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0129] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 6 not shown, usually called a "hard drive"). Although Figure 6 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical medium) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present application.

[0130] A program / utility 40 having a set (at least one) of program modules 46 may be stored, for example, in the memory 28, such program modules 46 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 46 generally perform the functions and / or methods of the embodiments described herein.

[0131] The electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the electronic device 12, and / or may communicate with any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). Such communication may be performed via an input / output (I / O) interface 22. Furthermore, the electronic device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 via the bus 18. It should be understood that although Figure 6 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0132] The processing unit 16 executes various functional applications and data processing by running the program stored in the system memory 28, for example, a post-loan fund monitoring method provided in an embodiment of the present invention is implemented: the consumption characteristic information of the target loan user is obtained through the blockchain; the consumption characteristic index parameters are determined according to the consumption characteristic information and a pre-trained characteristic analysis model; wherein the characteristic analysis model is trained with a gradient boosting decision tree as the original model and with the consumption characteristic information of historical loan users as input data; the post-loan risk factor of the target loan user is determined according to the consumption characteristic index parameters and a pre-trained risk assessment model, and the post-loan funds of the target loan user are monitored based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and with the consumption characteristic index parameters of historical loan users as input data.

[0133] The embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, a monitoring method for post-loan funds provided by all the embodiments of the present invention is implemented: obtaining consumption feature information of a target loan user through a blockchain; determining consumption feature index parameters according to the consumption feature information and a pre-trained feature analysis model; wherein the feature analysis model is trained with a gradient boosting decision tree as the original model and with consumption feature information of historical loan users as input data; determining the post-loan risk factor of the target loan user according to the consumption feature index parameters and a pre-trained risk assessment model, and monitoring the post-loan funds of the target loan user based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and with consumption feature index parameters of historical loan users as input data. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electronic device, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction-executing electronic device, apparatus, or device.

[0134] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction-executing electronic device, apparatus, or device.

[0135] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0136] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0137] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A post-loan fund monitoring method, characterized in that: The method comprises: Obtain consumption characteristic information of target loan users through blockchain; Determine consumption characteristic index parameters according to the consumption characteristic information and a pre-trained characteristic analysis model; wherein the characteristic analysis model is obtained by training with a gradient boosting decision tree as the original model and with consumption characteristic information of historical loan users as input data; According to the consumption characteristic indicator parameters and a pre-trained risk assessment model, the post-loan risk factor of the target loan user is determined, and the post-loan funds of the target loan user are monitored based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and the consumption characteristic indicator parameters of historical loan users as input data.

2. The method according to claim 1, characterized in that Obtain consumption characteristic information of target loan users through blockchain, including: Obtaining original consumption information of the target loan user through blockchain; wherein the original consumption information includes basic consumption information and structured information; Determining first consumption characteristic information and second consumption characteristic information according to the structured information; and determining third consumption characteristic information according to the basic consumption information; The first consumption characteristic information, the second consumption characteristic information and the third consumption characteristic information are determined as the consumption characteristic information.

3. The method according to claim 2, characterized in that The structured information includes at least one substructure information; and determining the second consumption characteristic information according to the structured information includes: Determine, according to each type of the substructure information and the information type of the substructure information, an information type field corresponding to each type of the substructure information; The information type fields corresponding to all the substructure information are determined as the second consumption characteristic information.

4. The method according to claim 1, characterized in that: The method for training the feature analysis model comprises: Acquire a first training set; wherein the first training set includes consumption characteristic information of at least one of the historical loan users, and one consumption characteristic information of the historical loan user corresponds to a first label information, and the first label information is used to indicate the importance of the consumption characteristic information of the historical loan user; Inputting the consumption characteristic information of the current historical loan user and the first label information corresponding to the consumption characteristic information of the current historical loan user into the gradient boosting decision tree model to obtain the first actual label information of the current historical loan user, wherein the first actual label information is used to indicate the importance of the consumption characteristic information of the historical loan user output by the gradient boosting decision tree model; If the error between the first actual label information of the current historical loan user and the first label information is less than or equal to a first preset threshold, determining the gradient boosting decision tree model as the feature analysis model; If the error between the first actual label information of the current historical loan user and the first label information is greater than a first preset threshold, the model parameters of the gradient boosting decision tree model are adjusted, the next historical loan user is taken as the current historical loan user, and the step of inputting the consumption feature information of the current historical loan user and the label information of the current historical loan user into the gradient boosting decision tree model is returned to execute to obtain the first actual label information of the current historical loan user.

5. The method according to claim 4, characterized in that Adjusting the model parameters of the gradient boosting decision tree model includes: Determining current parameter combination data of the gradient boosting decision tree model; The current parameter combination data is optimized using hyperparameter optimization to obtain optimized current parameter combination data.

6. The method according to claim 4, characterized in that The method of training the risk assessment model comprises: Acquire a second training set; wherein the second training set includes a consumption characteristic indicator parameter of at least one historical loan user output by the characteristic analysis model, and one consumption characteristic indicator parameter of the historical loan user corresponds to a second label information, and the second label information is used to indicate a post-loan risk indicator of the historical loan user; Inputting the consumption characteristic index parameter of the current historical loan user and the second label information corresponding to the consumption characteristic index parameter of the current historical loan user into the logistic regression model to obtain the second actual label information of the current historical loan user, wherein the second actual label information is used to indicate the post-loan risk index of the historical loan user output by the logistic regression model; If the error between the second actual label information of the current historical loan user and the second label information is less than or equal to a second preset threshold, determining the logistic regression model as the risk assessment model; If the error between the second actual label information of the current historical loan user and the second label information is greater than the second preset threshold, the model parameters of the logistic regression model are adjusted, the next historical loan user is taken as the current historical loan user, and the step of returning to execute the step of inputting the consumption characteristic indicator parameters of the current historical loan user and the second label information corresponding to the consumption characteristic indicator parameters of the current historical loan user into the logistic regression model to obtain the second actual label information of the current historical loan user.

7. A post-loan fund monitoring device, characterized in that: The device comprises: an information acquisition module, a parameter determination module and a risk monitoring module; wherein, The information acquisition module is used to obtain consumption characteristic information of the target loan user through the blockchain; The parameter determination module is used to determine the consumption characteristic index parameters according to the consumption characteristic information and a pre-trained characteristic analysis model; wherein the characteristic analysis model is trained by using a gradient boosting decision tree as the original model and using the consumption characteristic information of historical loan users as input data; The risk monitoring module is used to determine the post-loan risk factor of the target loan user based on the consumption characteristic indicator parameters and a pre-trained risk assessment model, and monitor the post-loan funds of the target loan user based on the post-loan risk factor; wherein the risk assessment model is trained with a logistic regression model as the original model and consumption characteristic indicator parameters of historical loan users as input data.

8. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the post-loan fund monitoring method as described in any one of claims 1 to 6.

9. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, a post-loan fund monitoring method as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the computer program implements the post-loan fund monitoring method according to any one of claims 1 to 6.

Images