User classification method and device, electronic equipment and storage medium
By constructing and processing the supply chain-user sparse matrix, generating user feature vectors and performing classification processing, the problem of inaccurate user classification results in the prior art is solved, and a more accurate and objective user classification is achieved.
Patent Information
- Application Number
- CN202411769860.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-04
- Publication Date
- 2025-02-28
AI Technical Summary
In the prior art, the user classification results are not accurate enough to meet the needs of enterprises, especially when the supply chain financing scenarios are complex.
By building a supply chain-user sparse matrix, empowering and dimensionality reduction processing, users' feature vectors are generated, and input into pre-trained classifiers for processing to obtain more accurate user classification results.
It improves the accuracy and objectivity of user classification and can better meet the needs of enterprises for user classification.
Smart Images
Figure CN119538010A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of financial technology, and in particular to a method, device, electronic device and storage medium for user classification. Background Art
[0002] As market competition becomes increasingly fierce, differentiated services and personalized marketing have become important factors for financial institutions to improve their market competitiveness. User classification can help banks better understand users, identify potential market opportunities, and help banks develop personalized products and services for different users, thereby improving user loyalty and satisfaction.
[0003] In the existing technology, when enterprises or financial institutions classify users, they consider relatively simple financing scenarios and indicators. However, the supply chain financing scenario is complex, and users can finance and apply for multiple products in multiple business districts, which is a one-to-many consumption scenario. This also leads to inaccurate user classification results and cannot meet the needs of enterprises. Summary of the invention
[0004] In view of this, the present application provides a method, device, electronic device and storage medium for user classification to solve the problem that the user classification results in the prior art are not accurate enough.
[0005] To achieve the above objectives, this application provides the following technical solutions:
[0006] The first aspect of the present application discloses a method for user classification, comprising:
[0007] Constructing a supply chain-user sparse matrix of the user to be classified; wherein the supply chain-user sparse matrix is used to characterize the association relationship between the supply chain and the user;
[0008] Performing weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix;
[0009] Performing dimensionality reduction processing on the weighted supply chain-user sparse matrix to obtain a supply chain feature vector of the user to be classified;
[0010] Based on the supply chain feature vector and the user feature vector of the user to be classified, a feature vector of the user is calculated; wherein the user feature vector is generated according to the user information of the user to be classified;
[0011] The feature vector is input into a pre-built classifier for processing to obtain a classification result of the user to be classified; wherein the classifier is pre-trained based on the sample feature vector.
[0012] Optionally, in the above method, constructing a supply chain-user sparse matrix includes:
[0013] For each supply chain, generate an attribute vector of the user to be classified under the current supply chain; wherein the attribute vector is used to characterize the attribute of the user to be classified related to the current supply chain;
[0014] The supply chain-user sparse matrix is generated based on each supply chain and the attribute vector of the user to be classified under each supply chain.
[0015] Optionally, in the above method, performing weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix includes:
[0016] Determine the weight value of each supply chain, and generate a weight matrix based on the weight value of each supply chain;
[0017] The weight matrix is multiplied by the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix.
[0018] Optionally, in the above method, the step of performing dimensionality reduction processing on the weighted supply chain-user sparse matrix to obtain a user feature vector includes:
[0019] The principal component analysis algorithm is used to perform dimensionality reduction processing on the weighted supply chain-user sparse matrix to obtain the supply chain feature vector of the user to be classified.
[0020] Optionally, in the above method, the training process of the classifier includes:
[0021] Obtaining sample feature vectors and actual classification results of the sample feature vectors;
[0022] Inputting the sample feature vector into an initial classifier for operation to obtain a classification result of the sample feature vector;
[0023] Determining whether the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector;
[0024] If the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector, the training of the classifier is completed;
[0025] If the classification result of the sample feature vector is different from the actual classification result of the sample feature vector, an error function is calculated and the parameters of the classifier are adjusted using the error function until the classification result of the sample feature vector output is the same as the actual classification result of the sample feature vector, and the training of the classifier is completed.
[0026] The second aspect of the present application discloses a user classification device, comprising:
[0027] A construction unit, used to construct a supply chain-user sparse matrix of the user to be classified; wherein the supply chain-user sparse matrix is used to characterize the association relationship between the supply chain and the user;
[0028] A weighting unit, used for performing weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix;
[0029] A dimension reduction unit, used for performing dimension reduction processing on the weighted supply chain-user sparse matrix to obtain a supply chain feature vector of the user to be classified;
[0030] A calculation unit, configured to calculate a user feature vector based on the supply chain feature vector and the user feature vector of the user to be classified; wherein the user feature vector is generated according to user information of the user to be classified;
[0031] The classification unit is used to input the feature vector into a pre-built classifier for processing to obtain a classification result of the user to be classified; wherein the classifier is pre-trained based on the sample feature vector.
[0032] Optionally, the above device, the construction unit, includes:
[0033] A first generating subunit is used to generate, for each supply chain, an attribute vector of the user to be classified under the current supply chain; wherein the attribute vector is used to represent the attribute of the user to be classified related to the current supply chain;
[0034] The second generating subunit is used to generate the supply chain-user sparse matrix based on each supply chain and the attribute vector of the user to be classified under each supply chain.
[0035] Optionally, the weighting unit of the above device includes:
[0036] determining a subunit, for determining a weight value of each supply chain, and generating a weight matrix based on the weight value of each supply chain;
[0037] The calculation subunit is used to multiply the weight matrix with the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix.
[0038] The third aspect of the present application discloses an electronic device, comprising:
[0039] one or more processors;
[0040] a storage device having one or more programs stored thereon;
[0041] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of the first aspects of the present invention.
[0042] A fourth aspect of the present application discloses a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method described in any one of the first aspects of the present invention is implemented.
[0043] It can be seen from the above technical scheme that in a method for user classification provided by the present application, a supply chain-user sparse matrix of the user to be classified is first constructed. The supply chain-user sparse matrix is weighted to obtain a weighted supply chain-user sparse matrix. The weighted supply chain-user sparse matrix is subjected to dimensionality reduction processing to obtain a supply chain feature vector of the user to be classified. Based on the supply chain feature vector and the user feature vector of the user to be classified, the feature vector of the user is calculated. The feature vector is input into a pre-built classifier for processing to obtain a classification result of the user to be classified. It can be seen that when the method of the present application is used to classify users, not only the characteristics of the user itself are considered, but also the connection between the user and each supply chain is considered, which ensures that the user classification result is more objective and improves the accuracy of user classification. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.
[0045] Figure 1 A flowchart of a method for user classification disclosed in an embodiment of the present application;
[0046] Figure 2 A dimensionality reduction principle diagram of the principal component analysis algorithm disclosed in the embodiment of the present application;
[0047] Figure 3 A flowchart of an implementation of the training process of the classifier in step S105 disclosed in the embodiment of the present application;
[0048] Figure 4 A schematic diagram of a user classification device disclosed in an embodiment of the present application;
[0049] Figure 5 A schematic diagram of an electronic device disclosed in an embodiment of the present application. DETAILED DESCRIPTION
[0050] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0051] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.
[0052] Furthermore, in this document, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.
[0053] As can be seen from the background technology, in the existing technology, when enterprises or financial institutions classify users, they consider relatively simple financing scenarios and indicators, while the supply chain financing scenario is complex, and users can finance and apply for multiple products in multiple business districts, which is a one-to-many consumption scenario. This also leads to the inaccurate user classification results and cannot meet the needs of enterprises.
[0054] In view of this, the present application provides a method, device, electronic device and storage medium for user classification to solve the problem that the user classification results in the prior art are not accurate enough.
[0055] The present application embodiment provides a method for user classification, such as Figure 1 As shown, specifically including:
[0056] S101. Construct a supply chain-user sparse matrix of the user to be classified; wherein the supply chain-user sparse matrix is used to characterize the association relationship between the supply chain and the user.
[0057] It should be noted that the supply chain-user sparse matrix of the user to be classified is constructed according to the attributes of the user to be classified in each supply chain (such as the number of loans, average loan amount, bad debt rate, etc.). The supply chain-user sparse matrix is used to characterize the relationship between the supply chain and the user.
[0058] It should also be noted that for supply chain finance, it usually extends around a core enterprise A and is associated with N non-core enterprises. However, not all corporate users in a supply chain have the same value. In the same supply chain, companies with more loans, larger loan amounts, and fewer default records are users with higher value. Users on the contrary have lower value. For users in multiple business districts, an enterprise may appear in the relationship network of multiple supply chains at the same time. It may have high-quality loans in a certain supply chain, but perform mediocrely in other supply chains. If the value of the corporate user is assessed solely by a single supply chain, the same user may get completely opposite results in different supply chains.
[0059] On the other hand, the quality of different supply chains is also different. For example, if there is a preferential interest rate policy for certain regions during a period of time, the loan amount and loan frequency of the user in the business district may increase significantly during this period; for example, due to the characteristics of certain industries, the loan amount of users in their business district is generally higher than that of other industries. The loan amount and loan frequency are important factors in judging the value of users, which will affect the correct classification of user value. Therefore, it is necessary to comprehensively consider the performance of a user in multiple supply chains and assign different weights to different supply chains to ensure that the user value results obtained are more accurate.
[0060] Optionally, in another embodiment of the present application, an implementation of the above step S101 may include:
[0061] For each supply chain, an attribute vector of the user to be classified under the current supply chain is generated; wherein the attribute vector is used to represent the attributes of the user to be classified related to the current supply chain.
[0062] Based on the attribute vectors of each supply chain and the user to be classified under each supply chain, a supply chain-user sparse matrix is generated.
[0063] It should be noted that for each user, the factors that affect their value classification can be divided into two parts. One part is the performance of the user in multiple supply chains, such as the user's loan amount, loan frequency, number of loans, number of overdue payments, etc.; the other part is the user's own information, such as the user's credit record, the user's tax record, etc., which reflect whether the user is high-quality. Here, only relevant information closely related to the user and the supply chain is considered.
[0064] For each supply chain, the relevant attributes of the user in the supply chain are listed to form an attribute vector m (loan amount, number of loans, loan frequency, number of overdue payments, ...) of the user to be classified in the current supply chain. Then, considering n supply chains, a two-dimensional matrix of the attribute vector of supply chain * user is formed, that is, a two-dimensional matrix of m dimensions * n dimensions, as shown below:
[0065]
[0066] Since users are not associated with every supply chain, this is a sparse matrix.
[0067] S102. Perform weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix.
[0068] It should be noted that when considering the user value classification, the quality of the supply chain itself will also have an impact on it. One is the impact of regional policies on supply chain users. For example, if a policy is introduced to implement loan interest rate discounts in a certain region during a period of time, then the supply chain business circle users in these regions may see a sharp increase in loan amount and loan frequency during this period, and after the interest rate discount ends, there may be a sharp drop in loan amount and loan frequency; in another supply chain where the core enterprise is unstable, due to the default of other non-core enterprise users, the user also has overdue loans and low loan quality scores; one is the impact of the scale of the supply chain on users. For example, if a user is in a supply chain with a hot industry as the core enterprise, due to the characteristics of the industry, the number of upstream and downstream enterprises in its supply chain is huge, the capital flow is large, and the loan amount is large. The direct reflection is that the user's loan amount in this supply chain is larger than in other supply chains, and the number of loans is more, which are all important factors affecting user value classification.
[0069] Therefore, the supply chain-user sparse matrix is weighted to obtain a weighted supply chain-user sparse matrix, so that the attributes of the supply chain with higher weights have a greater impact in the calculation, and the attributes of the supply chain with lower weights have a smaller impact in the calculation.
[0070] Optionally, in another embodiment of the present application, an implementation of the above step S102 may include:
[0071] A weight value of each supply chain is determined, and a weight matrix is generated based on the weight value of each supply chain.
[0072] Multiply the weight matrix by the supply chain-user sparse matrix to obtain the weighted supply chain-user sparse matrix.
[0073] It should be noted that when actually taking values, it is necessary to score the supply chain based on multiple factors such as supply chain stability, scale, and industry risk to determine the weight value of each supply chain. Then a weight matrix is generated based on the weight value of each supply chain. The weight matrix is multiplied by the supply chain-user sparse matrix to obtain the weighted supply chain-user sparse matrix, as shown below:
[0074]
[0075] in, Represents the weight of the i-th supply chain.
[0076] S103, performing dimensionality reduction processing on the weighted supply chain-user sparse matrix to obtain the supply chain feature vector of the user to be classified.
[0077] It should be noted that since users are not associated with all supply chains, the m*n matrix obtained above may contain a large amount of data that are all 0. This will not only waste storage space to store these meaningless data in the actual classification operation process, but also waste a lot of time for invalid operations on 0 elements.
[0078] Therefore, the weighted supply chain-user sparse matrix is subjected to dimensionality reduction processing to obtain the supply chain feature vector of the user to be classified. With minimal information loss, a feature vector containing the most key information is formed, while removing the redundant information in it, thus improving the efficiency and accuracy of the subsequent classification algorithm.
[0079] Optionally, in another embodiment of the present application, an implementation of step S103 may include:
[0080] The principal component analysis algorithm is used to reduce the dimension of the weighted supply chain-user sparse matrix to obtain the supply chain feature vector of the user to be classified.
[0081] It should be noted that in order to avoid the problem of data loss that may occur when directly reducing the dimensionality of the sparse matrix, the PCA principal component analysis algorithm is used to reduce the dimensionality of the weighted supply chain-user sparse matrix to obtain the supply chain feature vector of the user to be classified. This not only reduces data loss but also removes redundant data.
[0082] The principle of dimensionality reduction of principal component analysis algorithm is as follows Figure 2 As shown, taking the two-dimensional feature vector as an example, let Xi be a data point in the original coordinate space, that is, the position of the feature vector of user X in the supply chain i in the two-dimensional feature coordinates. PCA finds a vector e1 and projects the data point Xi onto the vector e1, making the distance between the data points before and after the projection as short as possible, ensuring that the data loss is as small as possible. That is:
[0083]
[0084] Among them, e represents the found vector, Xi is the original data point, Xi' is the projection data point of the original data point on the vector e, n represents the number of supply chains, and a vector e is found so that the sum of the distances of the feature data of all supply chains of the user after projection on the vector is the shortest. Vector e is the supply chain feature vector after dimensionality reduction.
[0085] S104. Based on the supply chain feature vector and the user feature vector of the user to be classified, a feature vector of the user is calculated; wherein the user feature vector is generated according to the user information of the user to be classified.
[0086] It should be noted that in addition to considering the supply chain feature vector of the user to be classified, it is also necessary to consider the user's own related characteristics, such as loan term, non-performing loan rate, user credit, total value of user assets, etc. The user's own information is formed into a one-dimensional user feature vector, for example, marked with vector l. The supply chain feature vector and the user feature vector are merged to obtain the user's feature vector. For example:
[0087] k=e+l
[0088] Among them, k is the user's feature vector, e is the supply chain feature vector, and l is the user feature vector.
[0089] S105, inputting the feature vector into a pre-built classifier for processing to obtain a classification result of the user to be classified; wherein the classifier is pre-trained according to the sample feature vector.
[0090] It should be noted that after calculating the feature vector of the user to be classified, the feature vector of the user to be classified is input into a pre-built classifier for processing to obtain the classification result of the user to be classified, and the classification result includes high-value users, medium-value users, and low-value users. The classifier is pre-trained based on the sample feature vector.
[0091] Optionally, in another embodiment of the present application, an implementation of the classifier training process in step S105 is as follows: Figure 3 As shown, specifically including:
[0092] S301, obtaining a sample feature vector and an actual classification result of the sample feature vector.
[0093] It should be noted that first, sample data of the user with annotated actual classification results is obtained, wherein the sample data includes sample feature vectors and the actual classification results annotated with each sample feature vector.
[0094] S302: Input the sample feature vector into an initial classifier for operation to obtain a classification result of the sample feature vector.
[0095] It should be noted that the sample feature vector is input into the initial classifier for operation to obtain the classification result of the sample feature vector. The type of classifier can be selected according to the actual situation, such as logistic regression, decision tree, support vector machine, naive Bayes, etc.
[0096] S303: Determine whether the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector.
[0097] S304: If the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector, the training of the classifier is completed.
[0098] It should be noted that if the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector, it means that the output result of the classifier is accurate, and the training of the classifier is completed.
[0099] S305. If the classification result of the sample feature vector is different from the actual classification result of the sample feature vector, an error function is calculated and the parameters of the classifier are adjusted using the error function until the classification result of the output sample feature vector is the same as the actual classification result of the sample feature vector, and the training of the classifier is completed.
[0100] It should be noted that if the classification result of the sample feature vector is different from the actual classification result of the sample feature vector, it means that the output result of the classifier is inaccurate. In this case, the error function is calculated and the parameters of the classifier are adjusted using the error function until the classification result of the output sample feature vector is the same as the actual classification result of the sample feature vector. The training of the classifier is then completed.
[0101] In a method for user classification provided in an embodiment of the present application, a supply chain-user sparse matrix of the user to be classified is first constructed. The supply chain-user sparse matrix is weighted to obtain a weighted supply chain-user sparse matrix. The weighted supply chain-user sparse matrix is subjected to dimensionality reduction processing to obtain a supply chain feature vector of the user to be classified. Based on the supply chain feature vector and the user feature vector of the user to be classified, the feature vector of the user is calculated. The feature vector is input into a pre-constructed classifier for processing to obtain a classification result of the user to be classified. It can be seen that when classifying users using the method of the present application, not only the characteristics of the user itself are considered, but also the connection between the user and each supply chain is considered, which ensures that the user classification result is more objective and improves the accuracy of user classification.
[0102] Another embodiment of the present application also discloses a user classification device, such as Figure 4 As shown, specifically including:
[0103] The construction unit 401 is used to construct a supply chain-user sparse matrix of the user to be classified; wherein the supply chain-user sparse matrix is used to represent the association relationship between the supply chain and the user.
[0104] The weighting unit 402 is used to perform weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix.
[0105] The dimension reduction unit 403 is used to perform dimension reduction processing on the weighted supply chain-user sparse matrix to obtain the supply chain feature vector of the user to be classified.
[0106] The calculation unit 404 is used to calculate the user's feature vector based on the supply chain feature vector and the user feature vector of the user to be classified; wherein the user feature vector is generated according to the user information of the user to be classified.
[0107] The classification unit 405 is used to input the feature vector into a pre-built classifier for processing to obtain a classification result of the user to be classified; wherein the classifier is pre-trained according to the sample feature vector.
[0108] In this embodiment, the specific execution process of the construction unit 401, the weighting unit 402, the dimension reduction unit 403, the calculation unit 404 and the classification unit 405 can be referred to the corresponding Figure 1 The content of the method embodiment will not be repeated here.
[0109] In a user classification device provided by the present application, first, a construction unit 401 constructs a supply chain-user sparse matrix of the user to be classified. The weighting unit 402 performs weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix. The dimension reduction unit 403 performs dimension reduction processing on the weighted supply chain-user sparse matrix to obtain a supply chain feature vector of the user to be classified. The calculation unit 404 calculates the user's feature vector based on the supply chain feature vector and the user feature vector of the user to be classified. The classification unit 405 inputs the feature vector into a pre-constructed classifier for processing to obtain a classification result of the user to be classified. It can be seen that when the method of the present application is used to classify users, not only the characteristics of the user itself are considered, but also the connection between the user and each supply chain is considered, which ensures that the user classification result is more objective and improves the accuracy of user classification.
[0110] Optionally, in another embodiment of the present application, an implementation of the construction unit 401 includes:
[0111] The first generating subunit is used to generate an attribute vector of the user to be classified under the current supply chain for each supply chain; wherein the attribute vector is used to represent the attributes of the user to be classified related to the current supply chain.
[0112] The second generating subunit is used to generate a supply chain-user sparse matrix based on the attribute vectors of each supply chain and the user to be classified under each supply chain.
[0113] In this embodiment, the specific execution process of the first generating subunit and the second generating subunit can be found in the corresponding method embodiment described above, and will not be repeated here.
[0114] Optionally, in another embodiment of the present application, an implementation of the weighting unit 402 includes:
[0115] A determination subunit is used to determine a weight value of each supply chain and generate a weight matrix based on the weight value of each supply chain.
[0116] The calculation subunit is used to multiply the weight matrix with the supply chain-user sparse matrix to obtain the weighted supply chain-user sparse matrix.
[0117] In this embodiment, the specific execution process of the determination subunit and the calculation subunit can be found in the corresponding method embodiment described above, which will not be repeated here.
[0118] Another embodiment of the present application further provides an electronic device, such as Figure 5 As shown, specifically including:
[0119] One or more processors 501 .
[0120] The storage device 502 stores one or more programs.
[0121] When one or more programs are executed by one or more processors 501 , the one or more processors 501 implement any one of the methods in the above embodiments.
[0122] Another embodiment of the present application further provides a computer storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, any one of the methods in the above embodiments is implemented.
[0123] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.
[0124] Professionals may further 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 the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may 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.
[0125] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for user classification, characterized in that: include: Constructing a supply chain-user sparse matrix of the user to be classified; wherein the supply chain-user sparse matrix is used to characterize the association relationship between the supply chain and the user; Performing weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix; Performing dimensionality reduction processing on the weighted supply chain-user sparse matrix to obtain a supply chain feature vector of the user to be classified; Based on the supply chain feature vector and the user feature vector of the user to be classified, a feature vector of the user is calculated; wherein the user feature vector is generated according to the user information of the user to be classified; The feature vector is input into a pre-built classifier for processing to obtain a classification result of the user to be classified; wherein the classifier is pre-trained based on the sample feature vector.
2. The method according to claim 1, characterized in that The constructing of the supply chain-user sparse matrix includes: For each supply chain, generate an attribute vector of the user to be classified under the current supply chain; wherein the attribute vector is used to characterize the attribute of the user to be classified related to the current supply chain; The supply chain-user sparse matrix is generated based on each supply chain and the attribute vector of the user to be classified under each supply chain.
3. The method according to claim 1, characterized in that The weighting process is performed on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix, including: Determine the weight value of each supply chain, and generate a weight matrix based on the weight value of each supply chain; The weight matrix is multiplied by the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix.
4. The method according to claim 1, characterized in that: The dimensionality reduction process is performed on the weighted supply chain-user sparse matrix to obtain the user's feature vector, including: The principal component analysis algorithm is used to perform dimensionality reduction processing on the weighted supply chain-user sparse matrix to obtain the supply chain feature vector of the user to be classified.
5. The method according to claim 1, characterized in that The training process of the classifier includes: Obtaining sample feature vectors and actual classification results of the sample feature vectors; Inputting the sample feature vector into an initial classifier for operation to obtain a classification result of the sample feature vector; Determining whether the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector; If the classification result of the sample feature vector is the same as the actual classification result of the sample feature vector, the training of the classifier is completed; If the classification result of the sample feature vector is different from the actual classification result of the sample feature vector, an error function is calculated and the parameters of the classifier are adjusted using the error function until the classification result of the sample feature vector output is the same as the actual classification result of the sample feature vector, and the training of the classifier is completed.
6. A user classification device, characterized in that: include: A construction unit, used to construct a supply chain-user sparse matrix of the user to be classified; wherein the supply chain-user sparse matrix is used to characterize the association relationship between the supply chain and the user; A weighting unit, used for performing weighting processing on the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix; A dimension reduction unit, used for performing dimension reduction processing on the weighted supply chain-user sparse matrix to obtain a supply chain feature vector of the user to be classified; A calculation unit, configured to calculate a user feature vector based on the supply chain feature vector and the user feature vector of the user to be classified; wherein the user feature vector is generated according to user information of the user to be classified; The classification unit is used to input the feature vector into a pre-built classifier for processing to obtain a classification result of the user to be classified; wherein the classifier is pre-trained based on the sample feature vector.
7. The device according to claim 6, characterized in that The construction unit comprises: A first generating subunit is used to generate, for each supply chain, an attribute vector of the user to be classified under the current supply chain; wherein the attribute vector is used to represent the attribute of the user to be classified related to the current supply chain; The second generating subunit is used to generate the supply chain-user sparse matrix based on each supply chain and the attribute vector of the user to be classified under each supply chain.
8. The device according to claim 6, characterized in that The weighting unit comprises: determining a subunit, for determining a weight value of each supply chain, and generating a weight matrix based on the weight value of each supply chain; The calculation subunit is used to multiply the weight matrix with the supply chain-user sparse matrix to obtain a weighted supply chain-user sparse matrix.
9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 5.
10. A computer storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.