WO2021164376A1

WO2021164376A1 - Recommendation method, apparatus and device, and computer-readable storage medium

Info

Publication number: WO2021164376A1
Application number: PCT/CN2020/134031
Authority: WO
Inventors: 黄福华; 郑文琛; 刘畅
Original assignee: 深圳前海微众银行股份有限公司
Priority date: 2020-02-20
Filing date: 2020-12-04
Publication date: 2021-08-26
Also published as: CN111324813A

Abstract

A recommendation method, apparatus and device, and a computer-readable storage medium. The method comprises the following steps: upon receiving a recommendation request, acquiring recommendation party data corresponding to the recommendation request, and inputting the recommendation party data into a parameter model to obtain recommendation parameters corresponding to the recommendation party data (S100); sending the recommendation parameters to a federated data exchange component, such that the federated data exchange component acquires, from a plurality of data party terminals, data party parameters corresponding to the recommendation parameters, and feeds back the data party parameters (S200); and performing training by means of a recommendation module on the basis of the recommendation parameters and the data party parameters, so as to obtain a recommendation result corresponding to the recommendation request (S300).

Description

Recommended method, device, equipment and computer readable storage medium

This application claims the priority of the Chinese patent application filed on February 20, 2020, the application number is 202010107651.6, and the name is "Recommended methods, devices, equipment and computer-readable storage media", the entire contents of which are incorporated herein by reference. Applying.

Technical field

This application relates to the field of data processing, and in particular to a recommendation method, device, device, and computer-readable storage medium.

Background technique

At present, as the scale of e-commerce continues to expand, the number and types of products are growing rapidly, and customers need to spend a lot of time to find the products they want to buy. This process of browsing a large amount of irrelevant information and products will undoubtedly cause the loss of consumers who are submerged in the problem of information overload. In order to solve these problems, a personalized recommendation system came into being.

Personalized recommendation is to recommend information and products that users are interested in based on the user's interest characteristics and purchase behavior. The personalized recommendation system provides an ability to recommend items based on user data. This ability depends on user data. The larger the number of samples in the data set, the more features of each sample. Under other conditions unchanged, the better the performance of the model that can be trained, the more likely the results of the recommendation will be by the user. accept.

However, because recommendation service providers generally use their own user data to build a personalized recommendation system, the data richness of a single party is limited, and the trained model cannot achieve the best performance, resulting in low recommendation performance.

The above content is only used to assist the understanding of the technical solution of the application, and does not mean that the above content is recognized as prior art.

Technical solutions

The main purpose of this application is to provide a recommendation method, device, equipment, and computer-readable storage medium, aiming to solve the technical problem of low recommendation performance of a personalized recommendation system using unilateral data.

In order to achieve the above objective, this application provides a recommendation method, which includes the following steps:

When a recommendation request is received, acquiring recommender data corresponding to the recommendation request, inputting the recommender data into a parameter model, and obtaining recommendation parameters corresponding to the recommender data;

Sending the recommended parameter to the federal data exchange component, so that the federal data exchange component obtains the data party parameter corresponding to the recommended parameter at multiple data party terminals, and feeds back the data party parameter;

Based on the recommendation parameter and the data party parameter, training is performed by a recommendation module to obtain a recommendation result corresponding to the recommendation request.

In addition, in order to achieve the foregoing objective, the present application also provides a recommendation device, the recommendation device including:

The obtaining module is configured to obtain the recommender data corresponding to the recommendation request when the recommendation request is received, obtain the recommender data corresponding to the recommendation request, and input the recommender data into the parameter model to obtain the recommender data Corresponding recommended parameters;

A sending module, configured to send the recommended parameter to the federated data exchange component, so that the federated data exchange component obtains the data party parameter corresponding to the recommended parameter from multiple data party terminals, and feeds back the data party parameter;

The recommendation module is configured to perform training through the recommendation module based on the recommendation parameter and the data party parameter to obtain the recommendation result corresponding to the recommendation request.

In addition, in order to achieve the above object, the present application also provides a recommendation device, the recommendation device including: a memory, a processor, and a recommendation program stored in the memory and capable of running on the processor, the recommendation program When executed by the processor, the steps of the above-mentioned recommendation method are realized.

In addition, in order to achieve the foregoing objective, the present application also provides a computer-readable storage medium having a recommendation program stored on the computer-readable storage medium, and the recommendation program is executed by a processor to implement the steps of the foregoing recommendation method.

In this application, when a recommendation request is received, the recommender data corresponding to the recommendation request is obtained, and the recommender data is input into the parameter model to obtain the recommended parameters corresponding to the recommender data; and then the recommended parameters are sent to the federation A data exchange component for the federal data exchange component to obtain data party parameters corresponding to the recommended parameters at multiple data party terminals, and feed back the data party parameters; and then based on the recommended parameters and the data party parameters, The recommendation module is trained to obtain the recommendation result corresponding to the recommendation request, and the recommendation is performed jointly with the data of the data party’s terminal, so that the number of features of the recommended data set is increased, thereby improving the recommendation performance. At the same time, by only transmitting Recommend (data party) parameters without transmitting the recommender data and data party data, and then realize data interaction under the premise of protecting the user privacy of the recommender terminal and the data party terminal, and improve the security of user data.

Description of the drawings

FIG. 1 is a schematic structural diagram of a recommended device of a hardware operating environment involved in a solution of an embodiment of the present application;

2 is a schematic flowchart of the first embodiment of the recommendation method of this application;

FIG. 3 is a schematic diagram of the architecture of a personalized recommendation system in an embodiment of this application;

FIG. 4 is a schematic diagram of the architecture of a personalized recommendation system in another embodiment of this application;

FIG. 5 is a schematic diagram of the architecture of a personalized recommendation system in another embodiment of the application.

FIG. 6 is a schematic diagram of the architecture of a personalized recommendation system in yet another embodiment of the application;

Fig. 7 is a schematic diagram of functional modules of the recommending device of this application.

Embodiments of the present invention

It should be understood that the specific embodiments described here are only used to explain the present application, and are not used to limit the present application.

As shown in Fig. 1, Fig. 1 is a schematic structural diagram of a recommended device of a hardware operating environment involved in a solution of an embodiment of the present application.

The recommended device in the embodiments of this application may be a PC, or a smart phone, a tablet computer, an e-book reader, MP3 (Moving Picture Experts Group Audio Layer) III. Moving Picture Experts Group Audio Layer IV (Moving Picture Experts Group Audio Layer IV) player, portable computer and other portable terminal equipment with display function.

As shown in FIG. 1, the recommended device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to implement connection and communication between these components. The user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a magnetic disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

In an embodiment, the recommended device may also include a camera, RF (Radio Frequency (radio frequency) circuits, sensors, audio circuits, WiFi modules, etc. Among them, sensors such as light sensors, motion sensors and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, where the ambient light sensor can adjust the brightness of the display screen according to the brightness of the ambient light. As a kind of motion sensor, the gravity acceleration sensor can detect the magnitude of acceleration in various directions (usually three-axis), and can detect the magnitude and direction of gravity when it is stationary. It can be used to identify the application of recommended device posture (such as horizontal and vertical screen switching, Related games, magnetometer attitude calibration), vibration recognition related functions (such as pedometer, percussion), etc.; of course, the recommended equipment can also be equipped with other sensors such as gyroscope, barometer, hygrometer, thermometer, infrared sensor, etc. No longer.

Those skilled in the art can understand that the recommended device structure shown in FIG. 1 does not constitute a limitation on the recommended device, and may include more or fewer components than shown in the figure, or a combination of certain components, or different component arrangements.

As shown in FIG. 1, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a recommended program.

In the recommended device shown in Figure 1, the network interface 1004 is mainly used to connect to the back-end server and communicate with the back-end server; the user interface 1003 is mainly used to connect to the client (user side) and communicate with the client; The device 1001 may be used to call the recommended program stored in the memory 1005.

In this embodiment, the recommendation device includes: a memory 1005, a processor 1001, and a recommendation program stored on the memory 1005 and running on the processor 1001, wherein the processor 1001 calls the recommendation program stored in the memory 1005 During the program, and perform the steps of the recommended method in each of the following embodiments.

This application also provides a recommendation method. Referring to FIG. 2, FIG. 2 is a schematic flowchart of the first embodiment of the recommendation method of this application.

The recommended methods include:

Step S100, when a recommendation request is received, obtain recommender data corresponding to the recommendation request, input the recommender data into a parameter model, and obtain recommendation parameters corresponding to the recommender data;

In this embodiment, the recommendation method can be applied to a cloud server or a recommender terminal. When the recommendation method is applied to a cloud server, the user triggers a recommendation request through the recommender terminal, and the recommender terminal sends the recommendation request to the cloud server. When the recommendation method is applied to the recommender terminal, the user can trigger the recommendation request through the recommender terminal.

When a recommendation request is received, the recommender data corresponding to the recommendation request is obtained, for example, the recommender data is carried in the recommendation request, the recommender data is obtained by parsing the recommendation request, and then the recommender data is input into the parameter model to obtain The recommended parameter corresponding to the recommender data, where the recommended parameter may apply for identification information of each data in the recommended data, for example, when the recommender data is multiple items, the recommended parameter may be the identification information of each item.

Step S200, sending the recommended parameter to the federated data exchange component, so that the federated data exchange component can obtain the data party parameter corresponding to the recommended parameter at multiple data party terminals, and feed back the data party parameter;

In this embodiment, when the recommended parameter is obtained, the recommended parameter is sent to the federated data exchange component. The federated data exchange component communicates with the recommender terminal (or cloud server) and the data party terminal, and the federated data exchange component communicates with the terminal of the recommending party (or cloud server) and the terminal of the data party according to the recommendation. The parameter obtains the data party parameter corresponding to the recommended parameter in multiple data party terminals, and feeds back the data party parameter.

In an embodiment, the step S200 includes:

Send the recommended parameters to the federal data exchange component, where the federal data exchange component forwards the recommended parameters to multiple data party terminals, and each of the data party terminals will query the data party data corresponding to the recommended parameters Input the parameter model, obtain the data party parameters corresponding to the data party data, and send the data party parameters to the federated data exchange component, and the federated data exchange component feeds back the received data party parameters.

In another embodiment, the step S200 includes:

Send the recommended parameters to the federal data exchange component, where the federal data exchange component determines the target data party terminal among multiple data party terminals based on the sample identification corresponding to the recommended parameter, and forwards the recommended parameters to all data party terminals. The target data party terminal is provided for the target data party terminal to feed back the data party parameter.

When the recommended parameters are obtained, the recommended parameters can be sent to the federal data exchange component, and the federal data exchange component obtains the data party parameters from multiple data party terminals according to the recommended parameters. For example, the federal data exchange component obtains the data identifier corresponding to the recommended parameter ( The identification information of each parameter in the recommended parameter), the data identification is sent to multiple data party terminals, and each data party terminal queries its own data according to the data identification to obtain the data party data corresponding to the recommended parameter, and input the data party data into the value parameter The model obtains the data party sub-parameters corresponding to the data party data, where the data identifier of the data party data is the same as the data identifier of the recommended parameter, and the federated data exchange component integrates the data party parameters according to the data party sub-parameters, or the federated data exchange The data identification of each data party terminal is stored in the component. The federal data exchange component determines the target data party terminal whose data identification includes the data identification of the recommender data in the data identification of the data party terminal, and sends the data identification of the recommender data to the target data party. Terminal, so that the target data party terminal feeds back the data party sub-data, and the federated data exchange component integrates the data party parameters according to the data party sub-parameters. Among them, the data party parameter includes the parameters in each recommended parameter and the corresponding ranking. When the recommender data is each item, the data party parameter includes the identification information of each item and the ranking corresponding to each identification information.

In step S300, training is performed by a recommendation module based on the recommendation parameter and the data party parameter to obtain a recommendation result corresponding to the recommendation request.

In this embodiment, when the data party parameters are obtained, based on the recommended parameters and the data party parameters, prediction is made through the recommendation model to obtain the recommendation result. Specifically, the recommendation model training is performed according to the data party parameters to obtain the post-training Then, based on the trained recommendation model, the recommended parameters are trained to obtain the recommendation result, and then the recommendation result is obtained based on the data of multiple parties, which improves the recommendation performance of the personalized recommendation system.

Wherein, the recommendation result may be a sorted list of each recommended parameter, or a preset number of parameters ranked first in the sorted list of each recommended parameter may be used as the sorted result.

Referring to Fig. 3, Fig. 3 is a schematic diagram of the architecture of a personalized recommendation system in an embodiment of the application. The architecture of the personalized recommendation system includes:

Federal data exchange component Fate Exchange, which is used for non-plain text data exchange between the recommender terminal (or cloud server) and the data party terminal;

The recommender terminal, the recommender terminal may include 1 to n, and n is a positive integer greater than 1. When the recommended method is applied to the recommender’s terminal, the recommender’s terminal is installed with the recommendation system corresponding to the recommended method of implementation; when the recommended method is applied to the cloud server, the cloud server is installed with the recommendation system corresponding to the recommended method of implementation, and the cloud server has A data interface for communication and connection with each recommender's terminal. In Figure 3, the recommendation method is applied to the recommender's terminal, and each recommender's terminal is installed with a recommendation system.

The recommendation system includes:

The Data module is used to store the unilateral data of the recommender, that is, the recommender data;

Fate Guest module, federated learning client component, used for joint training and joint prediction of recommendation algorithms;

Rec recommendation service, used to collect user data, call recommendation algorithms, and provide recommendation services to users;

Data party (data party terminal), which can contain 1 to n, where n is a positive integer greater than 1. Among them, each data party contains:

Data, storing the unilateral data of the data party;

Fate Host, the federated learning host-side component, is used for joint training and joint prediction of recommendation algorithms.

Referring to FIG. 4, FIG. 4 is a schematic diagram of the architecture of a personalized recommendation system in another embodiment of the application. In FIG. 4, the recommendation method is applied to a recommender terminal, and each recommender terminal is installed with a recommendation system. The personalized recommendation system architecture includes:

The recommender (terminal) includes: client layer, access layer, service layer, data layer and algorithm layer; among them,

The customer tier includes customers 1 to n, the customer tier is a customer system that requires a recommendation service, and n is a positive integer greater than 1, and the customer tier initiates a recommendation request.

The access layer includes: permission control: used to authenticate customers; data interface: used to receive data; recommendation interface: receive recommendation requests and return recommendation results;

Service layer: data processing: processing data; recommendation service: executing recommendation logic, including recall and sorting;

The algorithm layer includes a unilateral algorithm module and a federated algorithm module; among them,

Unilateral algorithm module: When the sample id of the recommender does not exist in the sample id of the data side, a unilateral algorithm needs to be trained. The unilateral algorithm module includes: data upload: upload data; task scheduling: scheduling algorithm training; Tensorflow: training algorithm, Including recall algorithm and sorting algorithm; Tf Serving: use algorithm to predict;

Federated algorithm module: When the sample id of the recommender exists in the sample id of the data side, train the federated recommendation algorithm and use the algorithm to make predictions, including: Data Access: data upload; Fate Guest: initiate model training; Fate Serving: use algorithms to make predictions; Fate Proxy: non-plaintext data exchange with Fate Exchange;

The data layer includes: DB1 to DBn: respectively used to store unilateral data (recommended party data).

The data side (terminal) includes the algorithm layer and the data layer, among which,

The algorithm layer includes: Data Access: data upload, data query; Fate Host: model training;

The data layer includes: DB: data storage; Data Service: Data service, interact with Data Access.

In this embodiment, the recommendation method is applied to a cloud server, that is, the recommendation system is deployed on the cloud server. The cloud server provides interfaces for multiple recommender terminals and receives recommendation requests sent by each recommender terminal. And when the recommendation result is obtained by prediction through the recommendation model, the recommender terminal corresponding to the recommendation request of the recommendation result is sent to complete the recommendation process.

Referring to Figure 5, the recommendation system is deployed on a certain cloud server. For example, the system in the irregular graphic area in Figure 5 provides recommendation services to the client (the recommender terminal) in the form of an interface; at this time, the recommender terminal needs The uploaded recommender data is the privacy protection data in an organization (cloud server). By combining with the data of the data party's terminal, the number of features of the data set that can be used for recommendation is increased, thereby improving the recommendation performance.

In this embodiment, referring to FIG. 6, the recommendation method is applied to the recommender terminal, and the recommendation system is deployed on each recommender terminal (client system), for example, the system in the irregular graphic area in FIG. 6; the recommender terminal uses non-plain text Send the recommender data to the federal data exchange component. At this time, the customer's data (recommendor data) does not exit the customer system, and only needs to interact with Fate Exchenge in a non-plain text way to improve the security of the recommender's data. The data union with the data party terminal increases the number of features of the data set that can be used for recommendation, thereby improving the recommendation performance.

Further, in an embodiment, the recommendation method further includes:

When the extended information is received, assign a data interface to the newly added terminal corresponding to the extended information based on the extended information, and establish a communication connection with the newly added terminal based on the first data interface;

Wherein, the extended information includes recommender extended information or data provider extended information, and the newly added terminal includes recommender terminal or recommender terminal.

Specifically, the data party (terminal) in the system framework corresponding to the recommended method in this implementation can be extended. When the federated data exchange component receives the data party extension information of the new data party, it is based on the data party extension information as the data. The party allocates the corresponding data interface, and establishes a communication connection between the federated data exchange component and the new data party based on the data interface, so that the more data parties, the more features, and the better the recommendation performance.

The recommender in the system framework corresponding to the recommendation method in this implementation can be extended. Specifically, when the federated data exchange component receives the recommender extension information of a new recommender, it assigns the corresponding recommender based on the recommender extension information The data interface is used to establish a communication connection between the federated data exchange component and the new recommender based on the data interface, so as to provide recommendation functions for multiple recommender terminals at the same time.

In the recommendation method proposed in this embodiment, when a recommendation request is received, the recommender data corresponding to the recommendation request is obtained, and the recommender data is input into the parameter model to obtain the recommended parameters corresponding to the recommender data; and then send The recommended parameters are sent to the federal data exchange component, so that the federal data exchange component obtains data party parameters corresponding to the recommended parameters from multiple data party terminals, and feeds back the data party parameters; and then based on the recommended parameters and The data party parameters are trained by the recommendation module to obtain the recommendation result corresponding to the recommendation request, and the recommendation is performed jointly with the data of the data party terminal, so that the number of features of the data set used for recommendation is increased, thereby improving the recommendation performance At the same time, by only transmitting recommended (data party) parameters without transmitting recommender data and data party data, data interaction is realized under the premise of protecting the user privacy of the recommender terminal and the data party terminal, and the security of user data is improved.

Based on the first embodiment, a second embodiment of the recommendation method of the present application is proposed. In this embodiment, step S300 includes:

Step S310: Perform recommendation model training based on the data party parameters to obtain a trained recommendation model;

Step S320: Based on the trained recommendation model, train the recommendation parameters to obtain the recommendation result.

In this embodiment, when the data party parameters are obtained, the recommended model training is performed through the data party parameters, that is, the data party parameters are input to the recommended model to obtain the trained recommendation model, and based on the trained recommendation model, the training Recommended parameters, the recommended parameters are input to the recommended model after training to obtain the recommended results, and the recommendation is performed jointly with the data of the data party's terminal, so that the number of features of the recommended data set is increased, thereby improving the recommendation performance.

In the recommendation method proposed in this embodiment, the recommended model is trained based on the parameters of the data party to obtain the trained recommendation model; then based on the trained recommendation model, the recommended parameters are trained to obtain the recommendation result. The data of the party terminal is jointly recommended, so that the feature number of the data set used for recommendation is increased, thereby improving the recommendation performance.

Based on the first embodiment, a third embodiment of the recommendation method of the present application is proposed. In this embodiment, after step S200, the method further includes:

Step S400, receiving the prompt information that the data party parameter does not exist and fed back by the federated data exchange component, where the prompt information is fed back when the target data party terminal does not exist among the multiple data party terminals of the federated data exchange component;

Step S500, input the recommended parameters into a recall model to obtain recall data, and input the recall data into a ranking model to obtain a trained ranking model;

Step S600: Input the recommended parameters into the trained ranking model to obtain the recommended results.

In this embodiment, the cloud server or the data party terminal needs to send the recommended parameters to the federated data exchange component. When the recommended parameters are obtained, the federated data exchange component obtains the sample identification corresponding to the recommended parameter, and determines whether the sample identification of the data party terminal is There is a sample identifier corresponding to the recommender data.

Specifically, after the federal data exchange component obtains the data identifier of the recommended parameter, it sends the data identifier to multiple data party terminals, and each data party terminal determines whether there is a data identifier of the recommender data in the data identifier of its own data, and feeds back the result to Federal data exchange component.

Alternatively, the data identification of each data party terminal is stored in the federated data exchange component, and the federated data exchange component determines whether the data identification of the data party terminal includes the target data party terminal of the data identification of the recommender data.

When the federated data exchange component determines that the sample identifier corresponding to the recommended parameter does not exist in the sample identifier of the data party terminal, it feeds back the prompt information, and the recommender terminal or cloud server receives the federated data exchange component feedback that there is no target data party terminal Prompt information.

Then, the recommended parameters are trained through the recall model, and the recommended parameters are input to the recall model to obtain recall data; the ranking model is trained based on the recall data, and the recall data is input to the ranking model to obtain the trained ranking model; based on the trained ranking model Predicting the recommended parameters, that is, inputting the recommended parameters into a trained ranking model to obtain a prediction result.

The recommendation method proposed in this embodiment is to receive the prompt information that the data party parameter does not exist in the federated data exchange component, wherein, when the target data party terminal does not exist among the multiple data party terminals of the federated data exchange component, the feedback Prompt information; then input the recommended parameters into the recall model to obtain recall data, and input the recall data into the ranking model to obtain the trained ranking model; then input the recommended parameters into the trained ranking model to obtain The recommendation result is to implement unilateral data recommendation when the data of the data party cannot be obtained, thereby improving user experience.

Based on the foregoing embodiments, a fourth embodiment of the recommendation method of the present application is proposed. In this embodiment, step S100 includes:

Step S110, when a recommendation request is received, obtain the recommender identity information corresponding to the recommendation request;

Step S120, verifying the identity information of the recommending party;

Step S130, when the identity information of the recommender is verified, obtain the recommender data corresponding to the recommendation request.

In this embodiment, when a recommendation request is received, the recommender identity information corresponding to the recommendation request is obtained; then the recommender identity information is verified, for example, whether the recommender identity information exists in the authorized user information, if If it exists, the verification is passed; by verifying the identity of the recommender, the security of the information recommendation is improved, and the security of the data of the data party is improved.

The recommendation method proposed in this embodiment obtains the recommender’s identity information corresponding to the recommendation request when the recommendation request is received; then verifies the recommender’s identity information; and then when the recommender’s identity information is verified , Obtain the recommender data corresponding to the recommendation request, and improve the security of the information recommendation by verifying the identity of the recommender, thereby enhancing the security of the data of the data party.

An embodiment of the present application also provides a recommendation device. Referring to FIG. 7, the recommendation device includes:

The obtaining module 100 is configured to, when a recommendation request is received, obtain recommender data corresponding to the recommendation request, obtain recommender data corresponding to the recommendation request, and input the recommender data into a parameter model to obtain the recommender Recommended parameters corresponding to the data;

The sending module 200 is configured to send the recommended parameters to the federal data exchange component, so that the federal data exchange component obtains the data party parameters corresponding to the recommended parameters from multiple data party terminals, and feeds back the data party parameters;

The recommendation module 300 is configured to perform training through the recommendation module based on the recommendation parameter and the data party parameter to obtain the recommendation result corresponding to the recommendation request.

In an embodiment, the sending module 200 is further used for:

Send the recommended parameters to the federal data exchange component, where the federal data exchange component forwards the recommended parameters to multiple data party terminals, and each of the data party terminals will query the data party data corresponding to the recommended parameters Input the parameter model, obtain the data party parameters corresponding to the data party data, and send the data party parameters to the federated data exchange component, and the federated data exchange component feeds back the received data party parameters;

In an embodiment, the sending module 200 is further used for:

Sending the recommended parameters to the federal data exchange component, wherein the federal data exchange component determines the target data party terminal among multiple data party terminals based on the sample identifier corresponding to the recommended parameter, and forwards the recommended parameter to all data party terminals The target data party terminal is provided for the target data party terminal to feed back the data party parameter.

In an embodiment, the recommending device further includes:

Receiving the prompt information that the data party parameter does not exist and is fed back by the federated data exchange component, wherein when there is no target data party terminal among the multiple data party terminals of the federal data exchange component, the prompt information is fed back;

Input the recommended parameters into a recall model to obtain recall data, and input the recall data into a ranking model to obtain a trained ranking model;

The recommended parameters are input into the trained ranking model to obtain the recommended results.

In an embodiment, the recommendation module 300 is also used to:

Perform recommendation model training based on the data party parameters to obtain a trained recommendation model;

Based on the trained recommendation model, the recommendation parameters are trained to obtain the recommendation result.

In an embodiment, the recommending device is also used to:

When receiving the extended information, the federated data exchange component allocates a data interface to the newly added terminal corresponding to the extended information based on the extended information, and establishes a communication connection with the newly added terminal based on the first data interface;

In an embodiment, the recommending device is also used to:

In an embodiment, the obtaining module 100 is further used for:

When receiving the recommendation request, obtain the recommender identity information corresponding to the recommendation request;

Verify the identity information of the recommender;

When the identity information of the recommender is verified, the recommender data corresponding to the recommendation request is acquired.

For the method executed by the above-mentioned program modules, please refer to the various embodiments of the recommended method of this application, which will not be repeated here.

In addition, the embodiment of the present application also proposes a computer-readable storage medium with a recommendation program stored on the computer-readable storage medium, and when the recommendation program is executed by a processor, the steps of the recommendation method as described above are implemented.

For the method implemented when the script calling program running on the processor is executed, please refer to the various embodiments of the recommended method of this application, which will not be repeated here.

It should be noted that in this article, the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements not only includes those elements, It also includes other elements that are not explicitly listed, or elements inherent to the process, method, article, or system. Without more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, method, article, or system that includes the element.

The serial numbers of the foregoing embodiments of the present application are for description only, and do not represent the superiority or inferiority of the embodiments.

Through the description of the above implementation manners, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is better.的实施方式。 Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM) as described above. , Magnetic disks, optical disks), including several instructions to make a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of the present application.

The above are only the preferred embodiments of the application, and do not limit the scope of the patent for this application. Any equivalent structure or equivalent process transformation made using the content of the description and drawings of the application, or directly or indirectly applied to other related technical fields , The same reason is included in the scope of patent protection of this application.

Claims

A recommendation method, wherein the recommendation method includes the following steps:

When a recommendation request is received, acquiring recommender data corresponding to the recommendation request, inputting the recommender data into a parameter model, and obtaining recommendation parameters corresponding to the recommender data;

Sending the recommended parameter to the federal data exchange component, so that the federal data exchange component obtains the data party parameter corresponding to the recommended parameter at multiple data party terminals, and feeds back the data party parameter;

Based on the recommendation parameter and the data party parameter, training is performed by a recommendation module to obtain a recommendation result corresponding to the recommendation request.
8. The recommendation method according to claim 1, wherein said sending said recommended parameters to a federated data exchange component, so that said federated data exchange component can obtain data party parameters corresponding to said recommended parameters from multiple data party terminals, And the step of feeding back the parameters of the data party includes:

Send the recommended parameters to the federal data exchange component, where the federal data exchange component forwards the recommended parameters to multiple data party terminals, and each of the data party terminals will query the data party data corresponding to the recommended parameters Input the parameter model, obtain the data party parameters corresponding to the data party data, and send the data party parameters to the federated data exchange component, and the federated data exchange component feeds back the received data party parameters.
3. The recommendation method according to claim 2, wherein said sending said recommended parameters to a federated data exchange component, wherein the step of said federated data exchange component forwarding said recommended parameters to multiple data party terminals comprises:

Sending the recommended parameters to the federal data exchange component, wherein the federal data exchange component determines the target data party terminal among multiple data party terminals based on the sample identifier corresponding to the recommended parameter, and forwards the recommended parameter to all data party terminals The target data party terminal is provided for the target data party terminal to feed back the data party parameter.
The recommendation method according to claim 3, wherein after the step of sending the recommendation parameter to the federated data exchange component, the method further comprises:

Receiving the prompt information that the data party parameter does not exist and is fed back by the federated data exchange component, wherein when there is no target data party terminal among the multiple data party terminals of the federal data exchange component, the prompt information is fed back;

Input the recommended parameters into a recall model to obtain recall data, and input the recall data into a ranking model to obtain a trained ranking model;

The recommended parameters are input into the trained ranking model to obtain the recommended results.
The recommendation method according to claim 1, wherein the step of training by a recommendation module based on the recommendation parameter and the data party parameter to obtain the recommendation list corresponding to the recommendation request comprises:

Perform recommendation model training based on the data party parameters to obtain a trained recommendation model;

Based on the trained recommendation model, the recommendation parameters are trained to obtain the recommendation result.
The recommendation method according to claim 1, wherein the recommendation method further comprises:

When receiving the extended information, the federated data exchange component allocates a data interface to the newly added terminal corresponding to the extended information based on the extended information, and establishes a communication connection with the newly added terminal based on the first data interface;

Wherein, the extended information includes recommender extended information or data provider extended information, and the newly added terminal includes recommender terminal or recommender terminal.
8. The recommendation method according to any one of claims 1 to 6, wherein when the recommendation request is received, the step of obtaining recommender data corresponding to the recommendation request comprises:

When receiving the recommendation request, obtain the recommender identity information corresponding to the recommendation request;

Verify the identity information of the recommender;

When the identity information of the recommender is verified, the recommender data corresponding to the recommendation request is acquired.
A recommendation device, wherein the recommendation device includes:

The obtaining module is configured to obtain the recommender data corresponding to the recommendation request when the recommendation request is received, obtain the recommender data corresponding to the recommendation request, and input the recommender data into the parameter model to obtain the recommender data Corresponding recommended parameters;

A sending module, configured to send the recommended parameter to the federated data exchange component, so that the federated data exchange component obtains the data party parameter corresponding to the recommended parameter from multiple data party terminals, and feeds back the data party parameter;

The recommendation module is configured to perform training through the recommendation module based on the recommendation parameter and the data party parameter to obtain the recommendation result corresponding to the recommendation request.
A recommendation device, wherein the recommendation device includes: a memory, a processor, and a recommendation program stored in the memory and capable of running on the processor, and the recommendation program is implemented as follows when the recommendation program is executed by the processor step:

When a recommendation request is received, acquiring recommender data corresponding to the recommendation request, inputting the recommender data into a parameter model, and obtaining recommendation parameters corresponding to the recommender data;

Sending the recommended parameter to the federal data exchange component, so that the federal data exchange component obtains the data party parameter corresponding to the recommended parameter at multiple data party terminals, and feeds back the data party parameter;

Based on the recommendation parameter and the data party parameter, training is performed by a recommendation module to obtain a recommendation result corresponding to the recommendation request.
9. The recommendation device according to claim 9, wherein said sending said recommended parameter to a federated data exchange component, so that said federated data exchange component can obtain data party parameters corresponding to said recommended parameters from multiple data party terminals, And the step of feeding back the parameters of the data party includes:

Send the recommended parameters to the federal data exchange component, where the federal data exchange component forwards the recommended parameters to multiple data party terminals, and each of the data party terminals will query the data party data corresponding to the recommended parameters Input the parameter model, obtain the data party parameters corresponding to the data party data, and send the data party parameters to the federated data exchange component, and the federated data exchange component feeds back the received data party parameters.
The recommendation device according to claim 10, wherein said sending said recommendation parameter to a federated data exchange component, wherein the step of said federated data exchange component forwarding said recommended parameter to multiple data party terminals comprises:

Sending the recommended parameters to the federal data exchange component, wherein the federal data exchange component determines the target data party terminal among multiple data party terminals based on the sample identifier corresponding to the recommended parameter, and forwards the recommended parameter to all data party terminals The target data party terminal is provided for the target data party terminal to feed back the data party parameter.
The recommendation device according to claim 11, wherein after the step of sending the recommendation parameter to a federated data exchange component, the method further comprises:

Receiving the prompt information that the data party parameter does not exist and is fed back by the federated data exchange component, wherein when there is no target data party terminal among the multiple data party terminals of the federal data exchange component, the prompt information is fed back;

Input the recommended parameters into a recall model to obtain recall data, and input the recall data into a ranking model to obtain a trained ranking model;

The recommended parameters are input into the trained ranking model to obtain the recommended results.
The recommendation device according to claim 9, wherein the step of training through a recommendation module based on the recommendation parameter and the data party parameter to obtain the recommendation list corresponding to the recommendation request comprises:

Perform recommendation model training based on the data party parameters to obtain a trained recommendation model;

Based on the trained recommendation model, the recommendation parameters are trained to obtain the recommendation result.
The recommendation device according to any one of claims 9 to 13, wherein when the recommendation request is received, the step of obtaining recommender data corresponding to the recommendation request comprises:

When receiving the recommendation request, obtain the recommender identity information corresponding to the recommendation request;

Verify the identity information of the recommender;

When the identity information of the recommender is verified, the recommender data corresponding to the recommendation request is acquired.
A computer-readable storage medium, wherein a recommended program is stored on the computer-readable storage medium, and the following steps are implemented when the recommended program is executed by a processor:

When a recommendation request is received, acquiring recommender data corresponding to the recommendation request, inputting the recommender data into a parameter model, and obtaining recommendation parameters corresponding to the recommender data;

Sending the recommended parameter to the federal data exchange component, so that the federal data exchange component obtains the data party parameter corresponding to the recommended parameter at multiple data party terminals, and feeds back the data party parameter;

Based on the recommendation parameter and the data party parameter, training is performed by a recommendation module to obtain a recommendation result corresponding to the recommendation request.
The computer-readable storage medium according to claim 15, wherein the sending the recommended parameter to a federal data exchange component, so that the federal data exchange component can obtain the data corresponding to the recommended parameter from a plurality of data party terminals The steps of feeding back the parameters of the data party include:

Send the recommended parameters to the federal data exchange component, where the federal data exchange component forwards the recommended parameters to multiple data party terminals, and each of the data party terminals will query the data party data corresponding to the recommended parameters Input the parameter model, obtain the data party parameters corresponding to the data party data, and send the data party parameters to the federated data exchange component, and the federated data exchange component feeds back the received data party parameters.
The computer-readable storage medium according to claim 16, wherein said sending said recommended parameter to a federated data exchange component, wherein the step of said federated data exchange component forwarding said recommended parameter to a plurality of data party terminals comprises :

Sending the recommended parameters to the federal data exchange component, wherein the federal data exchange component determines the target data party terminal among multiple data party terminals based on the sample identifier corresponding to the recommended parameter, and forwards the recommended parameter to all data party terminals The target data party terminal is provided for the target data party terminal to feed back the data party parameter.
17. The computer-readable storage medium according to claim 17, wherein after the step of sending the recommended parameters to a federal data exchange component, the method further comprises:

Receiving the prompt information that the data party parameter does not exist and is fed back by the federated data exchange component, wherein when there is no target data party terminal among the multiple data party terminals of the federal data exchange component, the prompt information is fed back;

Input the recommended parameters into a recall model to obtain recall data, and input the recall data into a ranking model to obtain a trained ranking model;

The recommended parameters are input into the trained ranking model to obtain the recommended results.
15. The computer-readable storage medium according to claim 15, wherein the step of training through a recommendation module based on the recommendation parameter and the data party parameter to obtain the recommendation list corresponding to the recommendation request comprises:

Perform recommendation model training based on the data party parameters to obtain a trained recommendation model;

Based on the trained recommendation model, the recommendation parameters are trained to obtain the recommendation result.
20. The computer-readable storage medium according to any one of claims 15 to 19, wherein when a recommendation request is received, the step of obtaining recommender data corresponding to the recommendation request comprises:

When receiving the recommendation request, obtain the recommender identity information corresponding to the recommendation request;

Verify the identity information of the recommender;

When the identity information of the recommender is verified, the recommender data corresponding to the recommendation request is acquired.