CN112214668A - A device and method for recommending personalized financial services based on big data - Google Patents

Abstract

The present invention discloses a big data-based personalized financial service recommendation device and method, comprising S1: acquiring user-service behavior statistical data, and establishing a user-service behavior statistical matrix T; S2: receiving an instruction and setting initialization parameters; S3: Construct a unified loss function Φ, and apply non-negative constraints to the unified loss function; S4: Use Nesterov accelerated gradient method to construct Nesterov accelerated user-service behavior recommendation model and train it; S5: Output user latent feature matrix and service latent feature matrix, for Restore the unknown behavior data in the user-service behavior statistics matrix. By adopting the Nesterov accelerated gradient method, the inherent statistical law of known user-service behavior statistical data is analyzed with a small computational complexity, thereby accurately restoring the unknown behavior data of personalized financial services based on big data, providing users with Personalized, precise financial services.

Description

A device and method for recommending personalized financial services based on big data

technical field

The present invention relates to the technical field of data processing, in particular to a device and method for recommending personalized financial services based on big data.

Background technique

With the continuous development of technologies such as big data and artificial intelligence, the traditional financial services with the structure of the industrial age are gradually moving towards scenario-based, personalized and intelligent personalized financial services, such as "Fixing Money", "Yiyibao", A large number of financial services such as "Yuebao" and "Yuebao Wealth Management" have emerged to fully satisfy users' consumption experience. However, with the continuous expansion of financial services, it is difficult for users to select the services they need from a large number of services. Therefore, the financial services industry hopes to actively capture users' consumption needs through big data technology and machine learning, and provide users with more accurate services .

According to the historical records of the financial service system, the user-service behavior statistical matrix is a commonly used data description structure, in which each row pair uses a user, each column corresponds to a service, and an element of each matrix corresponds to a user Historical behavioral data for a service. In financial services, users face a large number of services and cannot operate all services. Similarly, a service cannot be operated by all users, so the user-service behavior statistical matrix is often extremely sparse, and users have far more unknown behaviors about services. on known behavior.

In the financial service system, the behavior data of user services often have a natural law of positive numbers, so the financial service system adopts the existing non-negative matrix implicit feature analysis method to recommend services to users, but the non-negative matrix implicit feature analysis method has a faster convergence speed. It is slow and the data restoration accuracy is low, which makes it impossible to recommend the required services to users quickly and effectively. Therefore, how to perform efficient non-negative latent feature analysis for the extremely sparse user-service behavior statistical matrix in the financial service system, so as to be fast and accurate. Recommending required services for users and improving user experience is a key issue to achieve personalized finance.

SUMMARY OF THE INVENTION

Aiming at the problem of low matching degree between users and services in the prior art, the present invention proposes an apparatus and method for recommending personalized financial services based on big data. The recommendation accuracy and speed are improved, and the matching degree between users and services is improved.

In order to achieve the above object, the present invention provides the following technical solutions:

A personalized financial service recommendation device based on big data, comprising a data receiving module, a data storage module, an initialization module, a Nesterov training module and a service recommendation module; wherein,

The data receiving module is used to obtain user-service behavior statistical data from the financial service system and establish a user-service behavior statistical matrix, and store it in the data storage module;

The initialization module is used to preprocess the user-service behavior statistics matrix and initialize the parameters;

The Nesterov training module is used to construct a Nesterov accelerated gradient user-service behavior statistical model according to the user-service behavior statistics and the initialized parameters, so as to accelerate the recommendation user-service behavior statistics;

The service recommendation module is used to restore the user-service unknown behavior data through the constructed Nesterov accelerated gradient user-service behavior statistical model, and recommend corresponding services to the user according to the unknown behavior data.

Preferably, the user-service behavior statistics matrix is a matrix of |U| rows and |S| columns, where U represents a user set, and S represents a service set.

Preferably, the Nesterov training module includes a Nesterov update unit and a training unit; wherein,

The Nesterov update unit uses the Nesterov accelerated gradient method to update the parameters of the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C;

The training unit adopts the non-negative matrix latent feature analysis to train the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C.

Preferably, the service recommendation module includes an output hidden feature unit and a recommendation unknown interest degree unit; wherein,

Output latent feature unit, output user latent feature matrix X and service latent feature matrix Y in user-service behavior statistical data;

The unknown interest degree unit is recommended, and the user-service behavior data is restored according to the predicted user-service behavior statistical matrix T obtained from the user latent feature matrix X and the service latent feature matrix Y in the output user-service behavior statistical data.

The present invention also provides a method for recommending personalized financial services based on big data, which specifically includes the following steps:

S1: obtain user-service behavior statistical data, and establish a user-service behavior statistical matrix T;

S2: Receive commands and set initialization parameters;

S3: Construct a unified loss function and apply non-negative constraints to the unified loss function;

S4: The Nesterov accelerated user-service behavior recommendation model is constructed and trained using the Nesterov accelerated gradient method;

S5: Output the user latent feature matrix and the service latent feature matrix, predict and score the unknown behavior data in the user-service behavior statistical matrix, and recommend the service corresponding to the unknown behavior data of the top k to the user.

Preferably, the S1 includes:

The user set is U and the service set is S, and a matrix of |U| rows and |S| columns is established as the user-service behavior statistical matrix T, and then the reduction matrix factorization is used to decompose T to obtain the user latent feature matrix X respectively. and service latent feature matrix Y, where X is a matrix of |U| rows and f columns, each row vector in X corresponds to a user and is the hidden feature vector of the user; Y is a |S| row and f columns matrix, each row vector in Y corresponds to a service and is the hidden feature vector of the service; f is the dimension of the user implicit feature space and the service implicit feature space.

Preferably, the initialization parameters include latent feature matrices X and Y; latent feature space dimension f; Nesterov momentum control parameter μ, regularization factor λ; maximum number of iteration rounds R; Bias hidden eigenvectors B and C.

Preferably, in the S3, the expression of the unified loss function Φ(X, Y, B, C) is:

表示用户-服务行为统计矩阵中未知行为的还原值；b_u表示用户集合U的线性偏差第u行特征值；c_s表示服务集合S的线性偏差第s行特征值；x_u,k表示用户隐特征矩阵X的第u行第k列特征值；y_s,k表示服务隐特征矩阵Y的第s行第k列特征值；λ代表正则控制参数，f表示隐含特征空间的维数。In formula (1), t _{u, s} represents the entity relationship between user u and service s, that is, the historical behavior data of user u to service s; Λ represents the known behavior of users to financial services in the user-service behavior statistical matrix T data collection;

Represents the restored value of the unknown behavior in the user-service behavior statistical matrix; b _u represents the eigenvalue of the u-th row of the linear deviation of the user set U; c _s represents the s-th row eigenvalue of the linear deviation of the service set S; x _{u, k} represent the user The eigenvalue of the uth row and the kth column of the latent feature matrix X; y _s,k represents the eigenvalue of the sth row and the kth column of the service latent feature matrix Y; λ represents the regular control parameter, and f represents the dimension of the hidden feature space.

Preferably, the S4 includes the following steps:

S4-1: In the first iteration, the non-negative matrix latent feature analysis method is used to train the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C. The training formula is: :

In formula (2), |Λ(u)| represents the service set in the user-service known behavior data related to the user u, and |Λ(s)| represents the user-service known behavior data related to the service s. set of users;

S4-2: Use the Nesterov accelerated gradient method to update the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C. The update formula is:

表示b_u在t时刻的状态，

表示经过更新后的在t+1时刻b_u ^(t+1)的中间状态值，μ为Nesterov控制参数，控制加速效果，Nesterov加速梯度方法只在训练过程中第二次加入，即t≥2，表示迭代次数；In formula (3),

represents the state of b _u at time t,

Represents the updated intermediate state value of b _u ^(t+1) at time t+1, μ is the Nesterov control parameter, which controls the acceleration effect. The Nesterov acceleration gradient method is only added for the second time during the training process, that is, t≥2 , indicating the number of iterations;

进行训练，其训练公式为：S4-3: According to the non-negative matrix latent feature analysis method, the updated

For training, the training formula is:

表示使用中间状态

更新后在t+1时刻的状态值；parameters after training

Indicates the use of an intermediate state

The state value at time t+1 after the update;

S4-4: Determine whether the training reaches the iteration termination condition, if so, the iterative training ends, if not, continue training and updating until the iteration termination condition is reached.

Preferably, the S5 includes the following steps:

S5-1: Obtain hidden features corresponding to user u, service s, user linear deviation b and service linear deviation c from the output hidden feature unit;

即

S5-2: User u takes the inner product of the vector in the user latent feature matrix X and the service latent feature matrix Y corresponding to the service s and the user linear deviation b and the service linear deviation c as the unknown behavior of the user u to the service s data value

which is

To sum up, due to the adoption of the above technical solutions, compared with the prior art, the present invention has at least the following beneficial effects:

By adopting the Nesterov accelerated gradient method, the present invention analyzes the inherent statistical law of known user-service behavior statistical data with less computational complexity, thereby accurately restoring the unknown behavior data of personalized financial services based on big data, thereby Corresponding personalized and accurate financial services are provided for users; the present invention also realizes: (1) non-negative restriction is carried out on the user-service behavior statistical matrix through the non-negative matrix implicit feature analysis method, which improves the accuracy of recommendation, The user characteristics are better guaranteed; (2) the Nesterov accelerated gradient method is used to improve the recommendation accuracy; (3) the Nesterov accelerated gradient method is used to speed up the recommendation, save time and cost, and improve the matching degree between users and services.

Description of drawings:

FIG. 1 is a schematic diagram of an apparatus for recommending personalized financial services based on big data according to an exemplary embodiment of the present invention.

FIG. 2 is a schematic diagram of a method for recommending personalized financial services based on big data according to an exemplary embodiment of the present invention.

FIG. 3 is a schematic diagram of a time comparison diagram of user-service behavior statistical data analysis according to an exemplary embodiment of the present invention.

FIG. 4 is a schematic diagram of RMSE comparison according to an exemplary embodiment of the present invention.

Detailed ways

The present invention will be further described in detail below with reference to the examples and specific implementation manners. However, it should not be construed that the scope of the above-mentioned subject matter of the present invention is limited to the following embodiments, and all technologies realized based on the content of the present invention belong to the scope of the present invention.

In the description of the present invention, it should be understood that the terms "portrait", "horizontal", "upper", "lower", "front", "rear", "left", "right", "vertical", The orientations or positional relationships indicated by "horizontal", "top", "bottom", "inside", "outside", etc. are based on the orientations or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, rather than An indication or implication that the referred device or element must have a particular orientation, be constructed and operate in a particular orientation, is not to be construed as a limitation of the invention.

As shown in FIG. 1 , the present invention proposes a big data-based personalized financial service recommendation device, including a data receiving module 10 , a data storage module 20 , an initialization module 30 , a Nesterov training module 40 and a service recommendation module 50 . The output end of the data receiving module 10 is connected with the input end of the data storage module 20, the output end of the data storage module 20 is connected with the input end of the initialization module 30, the output end of the initialization module 30 is connected with the input end of the Nesterov training module 40, Nesterov The output end of the training module 40 is connected with the input end of the service recommendation module 50 . in,

The data receiving module 10 is used for acquiring user-service behavior statistical data from the financial service system and establishing a user-service behavior statistical matrix, which is stored in the data storage module 20 .

In the obtained user-service behavior statistics, the user set is denoted as U, and the service set is denoted as S, and a matrix with |U| rows and |S| columns is established as the user-service behavior statistical matrix T, and the user-service behavior The statistical matrix T is stored in the data storage module 20 . Then use the reduction matrix factorization to decompose T to obtain the user latent feature matrix X and the service latent feature matrix Y respectively. X is a matrix with |U| rows and f columns, and each row vector in X corresponds to a user, is the hidden feature vector of the user; Y is a matrix of |S| rows and f columns, each row vector in Y corresponds to a service and is the hidden feature vector of the service; f is the user’s hidden feature space and service The dimension of the hidden feature space; the user latent feature matrix X and the service latent feature matrix Y are stored in the data storage module 20 at the same time.

The initialization module 30 is used to initialize the parameters in the user-service behavior feature recommendation process.

The initialization parameters include the latent feature matrices X and Y of users and services; the hidden feature space dimension f; the Nesterov momentum control parameter μ, the regularization factor λ; the maximum number of iteration rounds R; the iteration round number control variable r in the training process; linear Bias hidden eigenvectors B and C. Among them, f determines the feature space dimension of each latent feature matrix, which is initialized as a positive integer; in the latent feature matrices X and Y: X is the latent feature matrix of |U| row and column f, and Y is the hidden feature matrix of |S| row and column f. The latent feature matrix is initialized with random small positive numbers respectively; in the linear deviation latent feature vectors B and C: B is the vector of |U| rows, C is the vector of |S| rows, and random small positive numbers are used respectively. The maximum number of training iterations R is a variable that controls the upper limit of the training number, and is initialized to a larger positive integer; the iteration number control variable r is initialized to 0; the regularization factor λ is a measure of the L ₂ regularization item to the recommended model The limiting effect of , initialized to a small positive number; Nesterov momentum control parameter μ, in order to control the acceleration effect of Nesterov momentum on the model, so that the recommended model can achieve the highest accuracy of data restoration of unknown behavior data, initialized to a positive number close to 1 .

The Nesterov training module 40 is configured to construct a Nesterov accelerated gradient user-service behavior statistical model according to the user-service behavior statistics and the initialized parameters, so as to accelerate the recommendation of the user-service behavior statistics.

In this embodiment, the Nesterov training module 40 includes a Nesterov updating unit and a training unit.

The Nesterov update unit uses the Nesterov accelerated gradient method to update the parameters of the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C, and update the corresponding hidden features in the form of a single element.

The training unit uses non-negative matrix latent feature analysis to train the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C, and update the corresponding latent features in the form of a single element.

The service recommendation module 50 is configured to restore the user-service unknown behavior data through the constructed Nesterov accelerated gradient user-service behavior statistical model, and output the user-service unknown behavior restoration data, and the hidden feature matrix of the user and the service.

In this embodiment, the service recommendation module 50 includes an output latent feature unit and a recommendation unknown interest degree unit.

Output latent feature unit, output user latent feature matrix X and service latent feature matrix Y in user-service behavior statistical data.

The unknown interest degree unit is recommended, and the user-service behavior data is restored according to the predicted user-service behavior statistical matrix T^ obtained from the user latent feature matrix X and the service latent feature matrix Y in the output user-service behavior statistical data.

Based on the above device, as shown in FIG. 2 , the present invention also proposes a method for recommending personalized financial services based on big data, which specifically includes the following steps:

S1: Obtain user-service behavior statistical data, and establish a user-service behavior statistical matrix T, for each element in the matrix, the row represents the corresponding user, and the column represents the corresponding service;

In the obtained user-service behavior statistics, the user set is denoted as U, and the service set is denoted as S, and a matrix with |U| rows and |S| columns is established as the user-service behavior statistical matrix T, and the user-service behavior The statistical matrix T is stored in the data storage module 20 . Then use the reduction matrix factorization to decompose T to obtain the user latent feature matrix X and the service latent feature matrix Y respectively. X is a matrix with |U| rows and f columns, and each row vector in X corresponds to a user, is the hidden feature vector of the user; Y is a matrix of |S| rows and f columns, each row vector in Y corresponds to a service and is the hidden feature vector of the service; f is the user’s hidden feature space and service The dimension of the implicit feature space.

S2: Receive commands and set initialization parameters;

In this embodiment, the initialized parameters include:

User-service known behavior data t _u,s , indicating that the entity relationship between user u and service s is the historical behavior data of user u to service s;

Linear deviation latent eigenvector B, representing the linear deviation eigenvector of user set U;

Linear deviation latent eigenvector C, representing the linear deviation eigenvector of service set S;

The latent feature matrix X represents the user latent feature matrix after factoring the user-service behavior statistical data matrix by the reduction matrix;

The latent feature matrix Y represents the service latent feature matrix after factoring the user-service behavior statistical data matrix by the reduction matrix;

The regular control parameter λ measures the limiting effect of the regular term on the model and enhances the robustness of the model;

Nesterov momentum control parameter μ, in order to control the acceleration effect of Nesterov momentum on the model, so that the data restoration accuracy of the recommended model for unknown behavior data can reach the highest level.

S3: Construct a unified loss function Φ(X, Y, B, C), and impose non-negative constraints on the unified loss function to ensure the non-negativity of the user and service latent feature matrices during the training process;

表示用户-服务行为统计矩阵中未知行为的还原值；b_u表示用户集合U的线性偏差第u行特征值；c_s表示服务集合S的线性偏差第s行特征值；x_u,k表示用户隐特征矩阵X的第u行第k列特征值；y_s,k表示服务隐特征矩阵Y的第s行第k列特征值；λ代表正则控制参数，f表示隐含特征空间的维数。In formula (1), t _{u, s} represents the entity relationship between user u and service s, that is, the historical behavior data of user u to service s; Λ represents the known behavior of users to financial services in the user-service behavior statistical matrix T data collection;

Represents the restored value of the unknown behavior in the user-service behavior statistical matrix; b _u represents the eigenvalue of the u-th row of the linear deviation of the user set U; c _s represents the s-th row eigenvalue of the linear deviation of the service set S; x _{u, k} represent the user The eigenvalue of the uth row and the kth column of the latent feature matrix X; y _s,k represents the eigenvalue of the sth row and the kth column of the service latent feature matrix Y; λ represents the regular control parameter, and f represents the dimension of the hidden feature space.

S4: The Nesterov accelerated user-service behavior recommendation model is constructed and trained using the Nesterov accelerated gradient method.

In this embodiment, the construction expression of the Nesterov accelerated user-service behavior recommendation model is represented by a unified loss function Φ(X, Y, B, C).

S4-1: In the first iteration, that is, t=1, use the non-negative matrix latent feature analysis method to train the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C , and its training formula is:

In formula (2), |Λ(u)| represents the service set in the user-service known behavior data related to the user u, and |Λ(s)| represents the user-service known behavior data related to the service s. collection of users.

S4-2: Add the Nesterov accelerated gradient method to accelerate the Nesterov accelerated user-service behavior recommendation model. First, use the Nesterov accelerated gradient method to calculate the user latent feature matrix X, the service latent feature matrix Y, the user linear deviation vector B and the service linear deviation vector C is updated, and its update formula is:

表示表示b_u在t时刻的状态，

表示表示经过更新后的在t+1时刻b_u ^(t+1)的中间状态值，μ为Nesterov控制参数，控制加速效果。Nesterov加速梯度方法只在训练过程中第二次加入，即t≥2，表示迭代次数。In formula (3),

represents the state of b _u at time t,

represents the updated intermediate state value of b _u ^(t+1) at time t+1, μ is the Nesterov control parameter, which controls the acceleration effect. The Nesterov accelerated gradient method is only added for the second time during the training process, that is, t≥2, which represents the number of iterations.

进行训练，其训练公式为：S4-3: According to the non-negative matrix latent feature analysis method, the updated

For training, the training formula is:

表示使用中间状态

更新后在t+1时刻的状态值。parameters after training

Indicates the use of an intermediate state

The state value at time t+1 after the update.

S4-4: Determine whether the training reaches the iteration termination condition, if so, the iterative training ends, if not, continue training and updating until the iteration termination condition is reached.

Iteration termination condition: if the iteration number control variable r exceeds the upper limit R of the training iteration number, the iterative training ends; if the iteration number control variable r is less than the upper limit R of the training iteration number (R can be 1000), continue training and updating , until the iteration termination condition is reached.

推荐服务，即未知行为数据

的值越大则表示用户使用该服务的几率也越大，由此将服务(取决于未知行为数据

的值)推荐给感兴趣的用户。S5: Output the user latent feature matrix and the service latent feature matrix, predict the unknown behavior data in the user-service behavior statistical matrix, and predict the unknown behavior data according to the predicted unknown behavior data.

Recommendation service, i.e. unknown behavioral data

The larger the value of , the higher the probability that the user will use the service, thus the service (depending on the unknown behavior data

value) are recommended to interested users.

S5-1: Obtain hidden features corresponding to user u, service s, user linear deviation b and service linear deviation c from the output hidden feature unit.

即

b_u表示用户集合U的线性偏差第u行特征值；c_s表示服务集合S的线性偏差第s行特征值；x_u,k表示用户隐特征矩阵X的第u行第k列特征值；y_s,k表示服务隐特征矩阵Y的第s行第k列特征值。S5-2: User u takes the inner product of the vector in the user latent feature matrix X and the service latent feature matrix Y corresponding to the service s and the user linear deviation b and the service linear deviation c as the unknown behavior of the user u to the service s data value

which is

b _u represents the eigenvalue of the _u -th row of the linear deviation of the user set U; c _s represents the s-th row of the eigenvalue of the linear deviation of the service set S; y _s,k represents the eigenvalue of the sth row and the kth column of the service latent feature matrix Y.

In this embodiment, FIG. 3 is a comparison diagram of the data analysis time of the recommendation model before and after applying the recommendation device and method of the present invention. As can be seen from FIG. 3 , after applying the recommending device and method of the present invention, the execution time for user-service behavior statistical data is much shorter than the case where the recommending device and method of the present invention are not applied. Specifically, it can be seen from FIG. 3 that after using the recommended device and method of the present invention, the execution time of the model without the recommended device and method of the present invention is about 7 times that of the recommended device and method of the present invention. That is, after using the recommending device and method of the present invention, the execution time of the model is increased by as much as 7 times, and the running efficiency of the model is greatly improved.

4 is a comparison diagram of RMSE in the data analysis process before and after applying the recommending device and method of the present invention. RMSE (root mean square error, root mean square error) is a measure of the recommending error, and the smaller the RMSE, the higher the accuracy. As can be seen from FIG. 4 , the application of the recommending device and method of the present invention greatly improves the precision of the user's personalized service selection in financial services. In practical applications, it can better serve users with personalized needs.

The big data-based personalized financial service recommendation device and method of the present invention adopts the Nesterov accelerated gradient method to analyze the inherent statistical law of the known user-service behavior statistical data with less computational complexity, Thus, the unknown behavior data of personalized financial services based on big data is accurately restored, and personalized and accurate financial services are provided for users; The matrix is non-negatively restricted to ensure the accuracy of the recommendation and better guarantee the user characteristics; (2) the Nesterov accelerated gradient method improves the recommendation accuracy of the recommendation model; (3) the Nesterov accelerated gradient method is used to speed up the The computation speed of the recommended model.

Those skilled in the art can understand that the above-mentioned embodiments are specific examples for realizing the present invention, and in practical applications, various changes in form and details can be made without departing from the spirit and the spirit of the present invention. scope.

Claims (10)

1. A personalized financial service recommendation device based on big data is characterized by comprising a data receiving module, a data storage module, an initialization module, a Nesterov training module and a service recommendation module; wherein,

the data receiving module is used for acquiring user-service behavior statistical data from the financial service system, establishing a user-service behavior statistical matrix and storing the user-service behavior statistical matrix in the data storage module;

the initialization module is used for preprocessing the user-service behavior statistical matrix and initializing parameters;

the Nesterov training module is used for constructing a Nesterov acceleration gradient user-service behavior statistical model according to the user-service behavior statistical data and the initialized parameters so as to accelerate and recommend the user-service behavior statistical data;

the service recommendation module is used for restoring unknown user-service behavior data through the constructed Nesterov acceleration gradient user-service behavior statistical model and recommending corresponding services to the user according to the unknown behavior data.

2. The apparatus as claimed in claim 1, wherein the statistical matrix of user-service behavior is a matrix of | U | rows and | S | columns, where U represents a user set and S represents a service set.

3. The big-data-based personalized financial service recommendation device of claim 1, wherein the Nesterov training module comprises a Nesterov update unit and a training unit; wherein,

a Nesterov updating unit which adopts a Nesterov acceleration gradient method to update parameters of the user hidden feature matrix X, the service hidden feature matrix Y, the user linear deviation vector B and the service linear deviation vector C;

and the training unit is used for training a user hidden feature matrix X, a service hidden feature matrix Y, a user linear deviation vector B and a service linear deviation vector C by adopting non-negative matrix hidden feature analysis.

4. The big-data-based personalized financial service recommendation device according to claim 1, wherein the service recommendation module comprises an output implicit feature unit and a recommendation unknown interestingness unit; wherein,

the output hidden feature unit outputs a user hidden feature matrix X and a service hidden feature matrix Y in the user-service behavior statistical data;

and the unknown interestingness recommending unit restores the unknown user-service behavior data according to the predicted user-service behavior statistical matrix T obtained by outputting the user implicit characteristic matrix X and the service implicit characteristic matrix Y in the user-service behavior statistical data.

5. A personalized financial service recommendation method based on big data is characterized by comprising the following steps:

s1, acquiring user-service behavior statistical data and establishing a user-service behavior statistical matrix T;

s2: receiving an instruction and setting initialization parameters;

s3: constructing a uniform loss function, and carrying out non-negative constraint on the uniform loss function;

s4: constructing a Nesterov acceleration user-service behavior recommendation model by adopting a Nesterov acceleration gradient method and training;

s5: and outputting the user hidden feature matrix and the service hidden feature matrix, performing predictive scoring on unknown behavior data in the user-service behavior statistical matrix, and recommending the service corresponding to the unknown behavior data of the top k to the user.

6. The method for recommending personalized financial services based on big data according to claim 5, wherein said S1 comprises:

the user set is U, the service set is S, a matrix of | U | rows and | S | columns is established as a user-service behavior statistical matrix T, the T is decomposed by using reduced matrix factorization to respectively obtain a user hidden feature matrix X and a service hidden feature matrix Y, wherein X is a matrix of | U | rows and f columns, and each row vector in X corresponds to one user and is a hidden feature vector of the user; y is a matrix of | S | rows and f columns, each row vector in Y corresponds to a service and is a hidden feature vector of the service; f is the dimension of the user implicit characteristic space and the service implicit characteristic space.

7. The method of claim 5, wherein the initialization parameters comprise hidden feature matrices X and Y; a hidden feature space dimension f; nesterov momentum control parameter mu and regularization factor lambda; the maximum iteration round number R; controlling a variable r by iteration rounds in the training process; the linear deviations hide the feature vectors B and C.

8. The method as claimed in claim 5, wherein in the S3, the uniform loss function Φ (X, Y, B, C) is expressed as:

in the formula (1), t_u,sRepresenting the entity relationship between the user u and the service s, namely historical behavior data of the user u to the service s; lambda represents a known behavior data set of the user to the financial service in the user-service behavior statistical matrix T;

representing a restoration value of an unknown behavior in the user-service behavior statistical matrix; b_uRepresenting the characteristic value of the U line of the linear deviation of the user set U; c. C_sThe characteristic value of the S line of the linear deviation representing the service set S; x is the number of_u,kRepresenting the characteristic value of the kth row and the kth column of the user hidden characteristic matrix X; y is_s,kRepresenting the characteristic value of the kth row and the kth column of the service hidden characteristic matrix Y; λ represents the canonical control parameter and f represents the dimension of the implicit feature space.

9. The method for recommending personalized financial services based on big data according to claim 5, wherein said S4 comprises the steps of:

s4-1: during the first iteration, a non-negative matrix hidden feature analysis method is adopted to train a user hidden feature matrix X, a service hidden feature matrix Y, a user linear deviation vector B and a service linear deviation vector C, and the training formula is as follows:

in formula (2), | Λ (u) | represents a service set in the user-service known behavior data related to the user u, and | Λ(s) | represents a user set in the user-service known behavior data related to the service s;

s4-2: updating a user hidden feature matrix X, a service hidden feature matrix Y, a user linear deviation vector B and a service linear deviation vector C by adopting a Nesterov acceleration gradient method, wherein the updating formula is as follows:

in the formula (3), the first and second groups,

denotes b_uIn the state at the time t, the state,

indicates updated at time b of t +1_u ^(t+1)The intermediate state value mu is a Nesterov control parameter and controls the acceleration effect, the Nesterov acceleration gradient method is only added for the second time in the training process, namely t is more than or equal to 2, and the iteration times are represented;

s4-3: after updating according to non-negative matrix hidden feature analysis method

Training is carried out, and the training formula is as follows:

trained parameters

Indicating the use of intermediate states

The state value at the moment t +1 after updating;

s4-4: and judging whether the training reaches an iteration termination condition, if so, finishing the iteration training, and if not, continuing the training and updating until the iteration termination condition is reached.

10. The method for recommending personalized financial services based on big data according to claim 5, wherein said S5 comprises the steps of:

s5-1: acquiring hidden features corresponding to a user u, a service s, a user linear deviation b and a service linear deviation c from an output hidden feature unit;

s5-2: the inner product of the vector of the user u to the user hidden feature matrix X and the vector of the service hidden feature matrix Y corresponding to the service s, the user linear deviation b and the service linear deviation c is used as the unknown behavior data value of the user u to the service s

Namely, it is

Images