Bank financing product characteristic classification standard generation method and system

Abstract

The invention discloses a bank financial product characteristic classification standard generation method and system, and is applied to the technical field of intelligent data processing, and the method comprises the steps: obtaining the product data of a plurality of financial product samples, and dividing the product data into product information data and product interaction data; classifying the financial product samples through the product information data to form a first classification, and classifying the financial product samples according to the product interaction data to form a second classification; mapping the first classification and the second classification into a standard feature space, and fusing the first classification and the second classification in the standard feature space to form a reference classification; and taking a classification condition corresponding to the reference classification as a reference classification standard. According to the technical scheme, the phenomenon that classification is inaccurate due to the fact that the financial product data dimension is too high is effectively reduced, and meanwhile due to the fact that more accurate classification is achieved, the generated classification category can serve as a standardized classification category to be used for subsequent product classification.

Description

Technical Field

[0001] The present invention relates to intelligent data processing technology, and in particular to a method and system for generating characteristic classification standards for bank financial products. Background Art

[0002] The classification of bank financial products refers to the classification and division of financial products issued by banks according to certain standards and characteristics, so that investors can better understand and choose products suitable for themselves. In the prior art, the classification of bank financial products is mainly to analyze the multi-dimensional data of financial products through cluster analysis, random forest model, neural network model, etc.; for example, the Chinese patent with application number 202310854863.4 discloses a classification method, device, equipment and readable storage medium for bank products, which can be applied to the fields of artificial intelligence and finance. The method includes: obtaining product information of the product to be classified; inputting the product information into a random forest classification model trained by the category influencing factors of bank products to obtain the classification result output by the random forest classification model; and outputting the classification result. However, in the actual classification process, if the relevant information and user preferences of bank financial products are comprehensively considered for classification, it is easy to have too high data dimensions, resulting in the non-convergence of the classification calculation process or the inability to meet the classification accuracy requirements. At this time, an accurate standardized classification result is required to facilitate the classification of subsequent products. Summary of the invention

[0003] In order to at least overcome the above-mentioned deficiencies in the prior art, the purpose of the present application is to provide a method and system for generating characteristic classification standards for bank financial products.

[0004] In a first aspect, an embodiment of the present application provides a method for generating a characteristic classification standard of a bank wealth management product, including: Acquire product data of multiple wealth management product samples, and divide the product data into product information data and product interaction data; the product information data includes minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the product interaction data includes user interaction data and corresponding user information; the user interaction data includes click data, purchase data and attention data; Classifying the wealth management product samples according to the product information data to form a first classification, and classifying the wealth management product samples according to the product interaction data to form a second classification; Mapping the first classification and the second classification into a standard feature space, and fusing the first classification and the second classification in the standard feature space to form a benchmark classification; The classification conditions corresponding to the benchmark classification are used as the benchmark classification standards.

[0005] When implementing the embodiment of the present application, it is first necessary to obtain the product data of the wealth management product sample, and its product data includes two types of data: one is the attribute data of the product itself, that is, the product information data, which needs to include data such as the minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the other is the data of the user's interaction with the wealth management product sample after it is put on the shelf, such as click data, purchase data and attention data, as well as the user's own information who interacts with the wealth management product sample, such as age, income, city of residence, etc. If all these data are aggregated to classify wealth management products, there will be a problem of high-dimensional data classification, and the final result will be unable to converge, overfitting, etc., so that an accurate classification category cannot be obtained. Therefore, in the embodiment of the present application, the wealth management product samples are classified in two different spaces, wherein the first space is the classification corresponding to the product information data, and the second space is the classification corresponding to the product interaction data; the different classifications generated after the two classifications can be fused in the standard feature space, which is a space containing the dimensions of all the above classifications. Since more accurate classification centers can be generated after fusion, reclassifying all wealth management product samples through these fixed classification centers can effectively reduce the clustering difficulties caused by high-dimensional data, thereby generating more accurate classification results, which can be used as standardized classification results for subsequent product classification. Through the above technical solution, the present application effectively reduces the inaccurate classification phenomenon caused by the excessively high dimensionality of wealth management product data, and since more accurate classification is achieved, the classification categories generated can be used as standardized classification categories for subsequent product classification.

[0006] In a possible implementation, classifying the wealth management product samples according to the product information data to form a first category includes: Constructing a first clustering space according to the product information data; the dimension of the first clustering space is the same as the dimension of the product information data; Randomly selecting a preset number of cluster centers in the first cluster space, and assigning the wealth management product samples to different cluster centers according to the product information data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the first classification.

[0007] In a possible implementation, classifying the wealth management product samples according to the product interaction data to form a second category includes: Normalizing the product interaction data to form normalized interaction data, and assigning the normalized interaction data to the wealth management product sample; constructing a second clustering space according to the normalized interaction data; the dimension of the second clustering space is the same as the dimension of the user information; Randomly selecting a preset number of cluster centers in the second cluster space, and assigning the wealth management product samples to different cluster centers according to the normalized interaction data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the second classification.

[0008] In a possible implementation, normalizing the product interaction data to form normalized interaction data includes: Assigning interaction weights to the click data, purchase data and attention data respectively, wherein the interaction weight of the purchase data is greater than the interaction weight of the attention data, and the interaction weight of the attention data is greater than the interaction weight of the click data; The normalized data of each wealth management product sample is calculated according to the following formula:

[0009] In the formula, is the interaction weight of the i-th user information of the j-th interacting user corresponding to the wealth management product sample; The i-th user information of the j-th interactive user corresponding to the financial product sample; is the i-th normalized data corresponding to the wealth management product sample.

[0010] In a possible implementation manner, mapping the first classification and the second classification into a standard feature space includes: Constructing the standard feature space; the number of dimensions of the standard feature space is the sum of the dimensions of the user information and the dimensions of the product information data; Obtaining a characteristic value of a classification center of the first classification, and obtaining a characteristic value of a classification center of the second classification; The classification center of the first classification and the classification center of the second classification are mapped into the standard feature space according to the feature values.

[0011] In a possible implementation, obtaining the characteristic value of the classification center of the first classification and obtaining the characteristic value of the classification center of the second classification includes: Acquire product information data corresponding to the classification center of the first classification as first classification product data, and acquire a weighted average of normalized data of financial product samples of each classification in the first classification as first classification user data; Obtaining normalized data corresponding to the classification center of the second classification as second classification user data, and obtaining a weighted average of product information data of the wealth management product samples of each classification in the second classification as second classification product data; the weighted weight of the weighted average is the normalized result of the inverse of the distance from the wealth management product sample to the corresponding classification center; The first classification product data and the first classification user data are used as feature values ​​of the classification center of the first classification; the second classification product data and the second classification user data are used as feature values ​​of the classification center of the second classification.

[0012] In a possible implementation manner, fusing the first classification and the second classification in the standard feature space to form a benchmark classification includes: Performing proximity classification on the classification center of the first classification and the classification center of the second classification in the standard feature space to form a plurality of classification categories, and taking the center of each of the classification categories as a benchmark clustering center; In the standard feature space, cluster analysis is performed on all the wealth management product samples with the benchmark cluster center as the cluster center to form a benchmark classification.

[0013] In the second aspect, the present application also provides a system for generating characteristic classification standards of bank wealth management products, including: an acquisition unit, configured to acquire product data of a plurality of wealth management product samples, and divide the product data into product information data and product interaction data; the product information data includes a minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the product interaction data includes user interaction data and corresponding user information; the user interaction data includes click data, purchase data and attention data; A first classification unit is configured to classify the wealth management product samples into a first classification according to the product information data, and to classify the wealth management product samples into a second classification according to the product interaction data; A second classification unit is configured to map the first classification and the second classification into a standard feature space, and fuse the first classification and the second classification in the standard feature space to form a reference classification; The standard unit is configured to use the classification condition corresponding to the reference classification as the reference classification standard.

[0014] In a possible implementation manner, the first classification unit is further configured to: Constructing a first clustering space according to the product information data; the dimension of the first clustering space is the same as the dimension of the product information data; Randomly selecting a preset number of cluster centers in the first cluster space, and assigning the wealth management product samples to different cluster centers according to the product information data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the first classification.

[0015] In a possible implementation manner, the first classification unit is further configured to: Normalizing the product interaction data to form normalized interaction data, and assigning the normalized interaction data to the wealth management product sample; constructing a second clustering space according to the normalized interaction data; the dimension of the second clustering space is the same as the dimension of the user information; Randomly selecting a preset number of cluster centers in the second cluster space, and assigning the wealth management product samples to different cluster centers according to the normalized interaction data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the second classification.

[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects: Through the above technical solution, the present invention effectively reduces the inaccurate classification phenomenon caused by the excessively high dimension of financial product data. At the same time, since a more accurate classification is achieved, the classification categories generated can be used as standardized classification categories for subsequent product classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings: Figure 1 This is a schematic diagram of the steps of the method of the embodiment of the present application; Figure 2 This is a schematic diagram of the fusion process of an embodiment of the present application. DETAILED DESCRIPTION

[0018] To make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme in the embodiment of the present application will be clearly and completely described below in conjunction with the drawings in the embodiment of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the embodiment of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0019] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0020] Please refer to Figure 1 , which is a flow chart of a method for generating characteristic classification standards of bank wealth management products provided in an embodiment of the present invention. The method for generating characteristic classification standards of bank wealth management products may specifically include the contents described in the following steps S1 to S4.

[0021] S1: Obtain product data of multiple wealth management product samples, and divide the product data into product information data and product interaction data; the product information data includes minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the product interaction data includes user interaction data and corresponding user information; the user interaction data includes click data, purchase data and attention data; S2: Classifying the wealth management product samples according to the product information data to form a first classification, and classifying the wealth management product samples according to the product interaction data to form a second classification; S3: Mapping the first classification and the second classification into a standard feature space, and fusing the first classification and the second classification in the standard feature space to form a benchmark classification; S4: The classification conditions corresponding to the benchmark classification are used as the benchmark classification standards.

[0022] When implementing the embodiment of the present application, it is first necessary to obtain the product data of the wealth management product sample, and its product data includes two types of data: one is the attribute data of the product itself, that is, the product information data, which needs to include data such as the minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the other is the data of the user's interaction with the wealth management product sample after it is put on the shelf, such as click data, purchase data and attention data, as well as the user's own information who interacts with the wealth management product sample, such as age, income, city of residence, etc. If all these data are aggregated to classify wealth management products, there will be a problem of high-dimensional data classification, and the final result will be unable to converge, overfitting, etc., so that an accurate classification category cannot be obtained. Therefore, in the embodiment of the present application, the wealth management product samples are classified in two different spaces, wherein the first space is the classification corresponding to the product information data, and the second space is the classification corresponding to the product interaction data; the different classifications generated after the two classifications can be fused in the standard feature space, which is a space containing the dimensions of all the above classifications. Since more accurate classification centers can be generated after fusion, reclassifying all wealth management product samples through these fixed classification centers can effectively reduce the clustering difficulties caused by high-dimensional data and generate more accurate classification results. These classification results can be used as standardized classification results for subsequent product classification. For example, each classification will correspond to a certain interval of product information data, which can be used as a classification condition to form a benchmark classification standard, or the data of each classification can be marked and then the neural network model can be trained to form a benchmark classification standard. Specifically, the benchmark classification is encoded to form a benchmark code; the product information data of the wealth management product samples in the benchmark classification is extracted, and a sample pair is formed with the corresponding benchmark code; the product information data in the sample pair is used as input data, and the benchmark code in the sample pair is used as output data to train the neural network model to generate a benchmark model; when a new wealth management product appears, the product information data of the wealth management product can be input into the benchmark model to generate a benchmark code corresponding to the benchmark classification to which the wealth management product should be classified. Through the above technical solution, the present application effectively reduces the phenomenon of inaccurate classification caused by the excessively high dimensionality of wealth management product data. At the same time, since a more accurate classification is achieved, the classification categories generated can be used as standardized classification categories for subsequent product classification.

[0023] In a possible implementation, classifying the wealth management product samples according to the product information data to form a first category includes: Constructing a first clustering space according to the product information data; the dimension of the first clustering space is the same as the dimension of the product information data; Randomly selecting a preset number of cluster centers in the first cluster space, and assigning the wealth management product samples to different cluster centers according to the product information data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the first classification.

[0024] When implementing the embodiment of the present application, the existing K-means mean clustering scheme is used to perform cluster analysis of the first classification. The dimensions of the product information data and the dimensions of the first clustering space are preferably 6, namely, the minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale. The number of preset cluster centers can be 10; in the first round of clustering, 10 cluster centers need to be randomly selected first, and the wealth management product samples are classified into these cluster centers. The classification process adopts the method of classifying the wealth management product samples to the cluster center closest to them, and the distance is preferably the Euclidean distance. After completing the first round of clustering, the new cluster center is recalculated and distributed again until the sum of the squares of the distances of all wealth management product samples to their cluster centers is minimized, and then the clustering is completed.

[0025] In a possible implementation, classifying the wealth management product samples according to the product interaction data to form a second category includes: Normalizing the product interaction data to form normalized interaction data, and assigning the normalized interaction data to the wealth management product sample; constructing a second clustering space according to the normalized interaction data; the dimension of the second clustering space is the same as the dimension of the user information; Randomly selecting a preset number of cluster centers in the second cluster space, and assigning the wealth management product samples to different cluster centers according to the normalized interaction data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the second classification.

[0026] When the embodiment of the present application is implemented, the same clustering process as the first classification is used to generate the second classification during the second classification process. The difference is that some characteristic normalization processing is required for the interactive user-related data to form normalized interactive data for classification.

[0027] In a possible implementation, normalizing the product interaction data to form normalized interaction data includes: Assigning interaction weights to the click data, purchase data and attention data respectively, wherein the interaction weight of the purchase data is greater than the interaction weight of the attention data, and the interaction weight of the attention data is greater than the interaction weight of the click data; The normalized data of each wealth management product sample is calculated according to the following formula:

[0028] In the formula, is the interaction weight of the i-th user information of the j-th interacting user corresponding to the wealth management product sample; The i-th user information of the j-th interactive user corresponding to the financial product sample; is the i-th normalized data corresponding to the wealth management product sample.

[0029] When the embodiment of the present application is implemented, each wealth management product sample may be browsed or purchased by a large number of users. At this time, normalizing these data becomes the most important content before classification. In the embodiment of the present application, different interaction weights are assigned to different interaction methods, which can effectively improve the accuracy of the normalized data. For example, please refer to Table 1, which shows the calculation method of normalized data in one case. In this example, user information includes age, city, number of children, annual income and years of work experience; the city is characterized by the average annual income of the city, and its unit is thousands of yuan; the interaction weight corresponding to the purchase data is selected as 4, the interaction weight of the click data is selected as 1, and the interaction weight of the attention data is selected as 2; in this example, users 1 to 8 are users who have interacted with a certain wealth management product.

[0030] Table 1 User information table

[0031] The following formula can be used to calculate each piece of user information corresponding to each wealth management product sample. After integrating all user information, the normalized data of the wealth management product sample is obtained: ; After calculation, the normalized data of the wealth management product sample is {42.47, 57.76, 0.88, 17.47, 19.11}, and the corresponding dimensions are age, city, number of children, annual income and years of work experience. It should be understood that the dimensions of the second clustering space of the second classification formed by cluster analysis should also be age, city, number of children, annual income and years of work experience.

[0032] In a possible implementation manner, mapping the first classification and the second classification into a standard feature space includes: Constructing the standard feature space; the number of dimensions of the standard feature space is the sum of the dimensions of the user information and the dimensions of the product information data; Obtaining a characteristic value of a classification center of the first classification, and obtaining a characteristic value of a classification center of the second classification; The classification center of the first classification and the classification center of the second classification are mapped into the standard feature space according to the feature values.

[0033] When the embodiment of the present application is implemented, the first category and the second category need to be mapped to the standard feature space to further find the most accurate classification center position. For the first category, it has clear product information data, but does not have clear normalized data; for the second category, it has clear normalized data, but does not have clear product information data; therefore, it is necessary to complete the above data through the process of obtaining characteristic values, and after completion, the mapping can be completed according to the corresponding data.

[0034] In a possible implementation, obtaining the characteristic value of the classification center of the first classification and obtaining the characteristic value of the classification center of the second classification includes: Acquire product information data corresponding to the classification center of the first classification as first classification product data, and acquire a weighted average of normalized data of financial product samples of each classification in the first classification as first classification user data; Obtaining normalized data corresponding to the classification center of the second classification as second classification user data, and obtaining a weighted average of product information data of the wealth management product samples of each classification in the second classification as second classification product data; the weighted weight of the weighted average is the normalized result of the inverse of the distance from the wealth management product sample to the corresponding classification center; The first classification product data and the first classification user data are used as feature values ​​of the classification center of the first classification; the second classification product data and the second classification user data are used as feature values ​​of the classification center of the second classification.

[0035] When the embodiment of the present application is implemented, the process of obtaining the characteristic value of the first classification and the second classification is similar, except that the data obtained is different. The normalized data required for the first classification needs to be obtained by weighted averaging the normalized data of all wealth management product samples corresponding to the classification center according to the distance. Similarly, the product information data required for the second classification is also obtained in this way.

[0036] In a possible implementation manner, fusing the first classification and the second classification in the standard feature space to form a benchmark classification includes: Performing proximity classification on the classification center of the first classification and the classification center of the second classification in the standard feature space to form a plurality of classification categories, and taking the center of each of the classification categories as a benchmark clustering center; In the standard feature space, cluster analysis is performed on all the wealth management product samples with the benchmark cluster center as the cluster center to form a benchmark classification.

[0037] When implementing the embodiments of this application, please refer to Figure 2 , shows the fusion process of the classification center of the first classification and the classification center of the second classification, wherein the dotted line represents the first classification, the solid line represents the second classification, the center of the dotted line represents the classification center of the first classification, the center of the solid line represents the classification center of the second classification, and the midpoint of the line connecting the center of the dotted line and the center of the solid line represents the benchmark cluster center. For example, the process of proximity classification can be adopted by each classification center of the first classification to find the classification center of the second classification closest to it to form a center point pair; the midpoint of the two points of the center point pair is calculated as the benchmark cluster center; the main reason is that the data sources of the first classification and the second classification are different, but their classification forms are relatively similar, which represents the connection between user preferences and product-related information. For example, in order to obtain the benchmark cluster center more accurately, the classification center of the first classification and the classification center of the second classification can also be clustered to form multiple cluster categories, and the cluster center of each cluster category is used as the benchmark cluster center, and the final result generated will be more accurate. After obtaining clear cluster centers, all financial product samples can be secondary clustered and classified in a high-dimensional standard feature space. Since the cluster centers are fixed, the clustering process transitions from an unsupervised classification process to a semi-supervised classification process. The calculation is easier to converge and the results produced will be more accurate.

[0038] In the second aspect, the present application also provides a system for generating characteristic classification standards of bank wealth management products, including: an acquisition unit, configured to acquire product data of a plurality of wealth management product samples, and divide the product data into product information data and product interaction data; the product information data includes a minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the product interaction data includes user interaction data and corresponding user information; the user interaction data includes click data, purchase data and attention data; A first classification unit is configured to classify the wealth management product samples into a first classification according to the product information data, and to classify the wealth management product samples into a second classification according to the product interaction data; A second classification unit is configured to map the first classification and the second classification into a standard feature space, and fuse the first classification and the second classification in the standard feature space to form a reference classification; The standard unit is configured to use the classification condition corresponding to the reference classification as the reference classification standard.

[0039] In a possible implementation manner, the first classification unit is further configured to: Constructing a first clustering space according to the product information data; the dimension of the first clustering space is the same as the dimension of the product information data; Randomly selecting a preset number of cluster centers in the first cluster space, and assigning the wealth management product samples to different cluster centers according to the product information data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the first classification.

[0040] In a possible implementation manner, the first classification unit is further configured to: Normalizing the product interaction data to form normalized interaction data, and assigning the normalized interaction data to the wealth management product sample; constructing a second clustering space according to the normalized interaction data; the dimension of the second clustering space is the same as the dimension of the user information; Randomly selecting a preset number of cluster centers in the second cluster space, and assigning the wealth management product samples to different cluster centers according to the normalized interaction data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the second classification.

[0041] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0042] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0043] The units described as separate components may or may not be physically separated. As units, it is obvious that a person of ordinary skill in the art can realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0044] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0045] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a grid device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0046] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for generating characteristic classification standards of bank financial products, characterized in that: include: Acquire product data of multiple wealth management product samples, and divide the product data into product information data and product interaction data; the product information data includes minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the product interaction data includes user interaction data and corresponding user information; the user interaction data includes click data, purchase data and attention data; Classifying the wealth management product samples according to the product information data to form a first classification, and classifying the wealth management product samples according to the product interaction data to form a second classification; Mapping the first classification and the second classification into a standard feature space, and fusing the first classification and the second classification in the standard feature space to form a benchmark classification; The classification conditions corresponding to the benchmark classification are used as the benchmark classification standards.

2. A method for generating characteristic classification standards of bank financial products according to claim 1, characterized in that: Classifying the wealth management product samples by using the product information data to form a first classification includes: Constructing a first clustering space according to the product information data; the dimension of the first clustering space is the same as the dimension of the product information data; Randomly selecting a preset number of cluster centers in the first cluster space, and assigning the wealth management product samples to different cluster centers according to the product information data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the first classification.

3. A method for generating characteristic classification standards of bank financial products according to claim 1, characterized in that: Classifying the wealth management product samples according to the product interaction data to form a second classification includes: Normalizing the product interaction data to form normalized interaction data, and assigning the normalized interaction data to the wealth management product sample; constructing a second clustering space according to the normalized interaction data; the dimension of the second clustering space is the same as the dimension of the user information; Randomly selecting a preset number of cluster centers in the second cluster space, and assigning the wealth management product samples to different cluster centers according to the normalized interaction data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the second classification.

4. A method for generating characteristic classification standards of bank financial products according to claim 3, characterized in that: Normalizing the product interaction data to form normalized interaction data includes: Assigning interaction weights to the click data, purchase data and attention data respectively, wherein the interaction weight of the purchase data is greater than the interaction weight of the attention data, and the interaction weight of the attention data is greater than the interaction weight of the click data; The normalized data of each wealth management product sample is calculated according to the following formula: In the formula, is the interaction weight of the i-th user information of the j-th interacting user corresponding to the wealth management product sample; The i-th user information of the j-th interactive user corresponding to the financial product sample; is the i-th normalized data corresponding to the wealth management product sample.

5. A method for generating characteristic classification standards of bank financial products according to claim 4, characterized in that: Mapping the first classification and the second classification into a standard feature space includes: Constructing the standard feature space; the number of dimensions of the standard feature space is the sum of the dimensions of the user information and the dimensions of the product information data; Obtaining a characteristic value of a classification center of the first classification, and obtaining a characteristic value of a classification center of the second classification; The classification center of the first classification and the classification center of the second classification are mapped into the standard feature space according to the feature values.

6. A method for generating characteristic classification standards of bank financial products according to claim 5, characterized in that: Obtaining the characteristic value of the classification center of the first classification and obtaining the characteristic value of the classification center of the second classification includes: Acquire product information data corresponding to the classification center of the first classification as first classification product data, and acquire a weighted average of normalized data of financial product samples of each classification in the first classification as first classification user data; Obtaining normalized data corresponding to the classification center of the second classification as second classification user data, and obtaining a weighted average of product information data of the wealth management product samples of each classification in the second classification as second classification product data; the weighted weight of the weighted average is the normalized result of the inverse of the distance from the wealth management product sample to the corresponding classification center; The first classification product data and the first classification user data are used as feature values ​​of the classification center of the first classification; the second classification product data and the second classification user data are used as feature values ​​of the classification center of the second classification.

7. A method for generating characteristic classification standards of bank financial products according to claim 5, characterized in that: Fusion of the first classification and the second classification in the standard feature space to form a benchmark classification includes: Performing proximity classification on the classification center of the first classification and the classification center of the second classification in the standard feature space to form a plurality of classification categories, and taking the center of each of the classification categories as a benchmark clustering center; In the standard feature space, cluster analysis is performed on all the wealth management product samples with the benchmark cluster center as the cluster center to form a benchmark classification.

8. A system for generating characteristic classification standards for bank financial products, characterized in that: include: an acquisition unit, configured to acquire product data of a plurality of wealth management product samples, and divide the product data into product information data and product interaction data; the product information data includes a minimum purchase amount, risk level, risk tolerance, product nature, investment period and investment scale; the product interaction data includes user interaction data and corresponding user information; the user interaction data includes click data, purchase data and attention data; A first classification unit is configured to classify the wealth management product samples into a first classification according to the product information data, and to classify the wealth management product samples into a second classification according to the product interaction data; A second classification unit is configured to map the first classification and the second classification into a standard feature space, and fuse the first classification and the second classification in the standard feature space to form a reference classification; The standard unit is configured to use the classification condition corresponding to the reference classification as the reference classification standard.

9. A bank financial product characteristic classification standard generation system according to claim 8, characterized in that: The first classification unit is further configured to: Constructing a first clustering space according to the product information data; the dimension of the first clustering space is the same as the dimension of the product information data; Randomly selecting a preset number of cluster centers in the first cluster space, and assigning the wealth management product samples to different cluster centers according to the product information data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the first classification.

10. A bank financial product characteristic classification standard generation system according to claim 8, characterized in that: The first classification unit is further configured to: Normalizing the product interaction data to form normalized interaction data, and assigning the normalized interaction data to the wealth management product sample; Constructing the second clustering space based on the normalized interaction data; The dimension of the second clustering space is the same as the dimension of the user information; Randomly selecting a preset number of cluster centers in the second cluster space, and assigning the wealth management product samples to different cluster centers according to the normalized interaction data; Recalculating the cluster center according to the allocation result, and re-allocating the wealth management product samples; Repeat the generation of the cluster center and the allocation of the financial product samples until the preset conditions are met, and use the current classification result as the second classification.

Images