Method and device for information recommendation, electronic device, and storage medium

Abstract

The present application relates to the technical field of information recommendation, and discloses a method for information recommendation, which is applied to a user device, comprising: when a new application is discovered, inputting the information to be recommended into a trained information recommendation model, and obtaining the probability of the user device clicking on the recommended information; and determining whether the information to be recommended is recommended in the application based on the probability of the user device clicking on the recommended information. By inputting the information to be recommended into a trained information recommendation model, the probability of the user device clicking on the recommended information is obtained, thereby achieving a feature characterization between the information to be recommended and the user device; based on the probability of the user device clicking on the recommended information obtained from the information recommendation model, determining whether the information to be recommended is recommended in the application, thereby making the information recommendation more accurate and the user experience better. The present application also discloses a device, an electronic device, and a storage medium for information recommendation.

Description

Technical Field

[0001] The present application relates to the technical field of information recommendation, for example, to a method and device, electronic device, and storage medium for information recommendation. Background Art

[0002] At present, with the development of Internet technology, various types of information appear on applications, such as product information, movie information, book information, news information, and video information. Users can search for the information they need through the search function in the application. At the same time, the application can also actively recommend information to users to help users find content of interest in massive data.

[0003] In the process of actively recommending information to the user side, the recommendation methods in the related technology include popularity-based recommendation, content-based recommendation or collaborative filtering recommendation, etc., all of which complete the recommendation purpose based on the historical behavior data of the user side, or discover users with similar or identical historical behavior data as the user, and then complete the recommendation purpose based on the historical behavior data; the related technology also provides an information recommendation method, including obtaining the user's historical information sequence and the recommendation information set; determining the correlation factor of each recommendation information in the recommendation information set corresponding to the historical information sequence; determining the user's behavior characteristics corresponding to each recommendation information based on the correlation factor of each recommendation information in the recommendation information set corresponding to the historical information sequence; performing multiple iterative feature extraction processes on the behavior characteristics of each recommendation information, and determining the click-through rate of each recommendation information based on the feature extraction results of each recommendation information; and performing a recommendation operation based on the click-through rate of each recommendation information.

[0004] During the implementation of the embodiments of the present disclosure, it was found that at least the following problems exist in the related art:

[0005] The recommendation method in the related art is relatively unpopular when the historical behavior data on the user side is relatively unpopular, and when there are very few users with similar historical behavior data as the user in the big data, the accuracy of information recommendation to the user is poor. Summary of the Invention

[0006] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical elements or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0007] The embodiments of the present disclosure provide a method and apparatus, an electronic device, and a storage medium for information recommendation, so as to more accurately recommend information to users.

[0008] In some embodiments, the method for information recommendation is applied to a user device, including: when a new application is discovered, the information to be recommended is input into a trained information recommendation model to obtain the probability of the user device clicking on the recommended information; based on the probability of the user device clicking on the recommended information, determining whether the information to be recommended is recommended in the application.

[0009] In some embodiments, the method for information recommendation is applied to a server, including: receiving n encrypted local model gradients sent by n user devices; wherein n is an integer greater than 1; calculating the average model gradient of the n encrypted local model gradients; and sending the average model gradient to each user device so that the n user devices obtain the probability of clicking on the information to be recommended, and determine whether the information to be recommended is recommended in the application.

[0010] In some embodiments, the device for information recommendation is provided on a user device and includes: a probability prediction module, configured to input the information to be recommended into a trained information recommendation model when a new application is discovered, to obtain the probability of the user device clicking on the recommended information; an information recommendation module, configured to determine whether the information to be recommended is recommended in the application based on the probability of the user device clicking on the recommended information.

[0011] In some embodiments, the device for information recommendation is provided on a server and includes: a gradient receiving module configured to receive n encrypted local model gradients sent by n user devices; wherein n is an integer greater than 1; a gradient calculation module configured to calculate the average model gradient of the n encrypted local model gradients; and a gradient sending module configured to send the average model gradient to each user device, so that the n user devices obtain the probability of clicking on the information to be recommended and determine whether the information to be recommended is recommended in the application.

[0012] In some embodiments, the electronic device includes a processor and a memory storing program instructions, and the processor is configured to execute the aforementioned method for information recommendation when running the program instructions.

[0013] In some embodiments, the storage medium stores program instructions, and when the program instructions are run, they execute the aforementioned method for information recommendation.

[0014] The method and apparatus, electronic device, and storage medium for information recommendation provided by the embodiments of the present disclosure can achieve the following technical effects:

[0015] By inputting the information to be recommended into a trained information recommendation model, the probability of the user device clicking on the recommended information is obtained, achieving a feature characterization between the information to be recommended and the user device; based on the probability of the user device clicking on the recommended information predicted by the information recommendation model, it is determined whether the information to be recommended is recommended in the application, thereby achieving personalized recommendations for each individual, without the need for excessive historical behavior data, making information recommendations more accurate and the user experience better. The method for information recommendation of the disclosed embodiment is used in the field of recommendation technology. Through the personalized recommendation method, it is well suited for cold start of applications and can also provide targeted recommendations to users.

[0016] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] One or more embodiments are exemplarily described by corresponding drawings. These exemplary descriptions and drawings do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements. The drawings do not constitute a scale limitation. In addition,

[0018] Figure 1 This is a schematic diagram of a federated learning architecture provided by an embodiment of the present disclosure;

[0019] Figure 2 is a schematic diagram of a method for information recommendation provided by an embodiment of the present disclosure;

[0020] Figure 3 is a schematic diagram of another method for information recommendation provided by an embodiment of the present disclosure;

[0021] Figure 4 is a schematic diagram of another method for information recommendation provided by an embodiment of the present disclosure;

[0022] Figure 5 is a schematic diagram of another method for information recommendation provided by an embodiment of the present disclosure;

[0023] Figure 6 is a schematic diagram of another method for information recommendation provided by an embodiment of the present disclosure;

[0024] Figure 7 is a schematic diagram of another method for information recommendation provided by an embodiment of the present disclosure;

[0025] Figure 8 is a schematic diagram of another method for information recommendation provided by an embodiment of the present disclosure;

[0026] Figure 9 This is a schematic diagram of the architecture of a method for information recommendation provided by an embodiment of the present disclosure;

[0027] Figure 10 is a schematic diagram of a device for information recommendation provided by an embodiment of the present disclosure;

[0028] Figure 11 is a schematic diagram of another device for information recommendation provided by an embodiment of the present disclosure;

[0029] Figure 12 2 is a schematic diagram of another device for information recommendation provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] In order to be able to understand the features and technical content of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The accompanying drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the sake of convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0031] In the description and claims of the embodiments of the present disclosure, as well as in the accompanying drawings, the terms "first," "second," and the like are used to distinguish similar items and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate to describe the embodiments of the present disclosure herein. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions.

[0032] Unless otherwise stated, the term "plurality" means two or more.

[0033] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B.

[0034] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0035] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.

[0036] Combine Figure 1The figure shows a schematic diagram of a horizontal federated learning architecture. K participants with the same data structure collaborate to train a machine learning model with the help of a manager (Server A). The manager is responsible for sending the AI ​​model to each distributed node and updating the model based on the feedback from the participants. The updated AI model is then sent back to the participants for the next round of model training. The database in the figure can be considered as multiple different participants, each with its own data. In order to avoid sending local data to others, such as the manager, the model sent by the manager is used for local training and then returned to the manager for model update. The participants can be network elements on the network side or terminal devices.

[0037] The model training process can include the following steps:

[0038] S1. Each participant downloads the latest model from server A;

[0039] S2. Each participant uses local data to train the model, and uploads the encrypted gradient to Server A. Server A aggregates the gradients of each user to update the model parameters.

[0040] S3: Server A returns the updated model to each participant;

[0041] S4: Each participant updates their own model.

[0042] The above steps S2 to S4 are repeated multiple times to update the model. Under certain conditions, such as when a certain number of iterations are completed or the calculated value of the model's loss function is lower than a preset value, the model training is completed.

[0043] This application is based on the training of the federated learning algorithm, which integrates the data of various applications on the user device and performs federated training locally on the user. It accurately characterizes the user device through the behavior of the user device in different applications. When a new information appears, the user device determines whether the information is displayed in the corresponding application based on the information recommendation model; or when a new application is started, the user device determines whether the new information is displayed in the new application based on the information recommendation model.

[0044] Combine Figure 2 As shown, an embodiment of the present disclosure provides a method for information recommendation, which is applied to a user device and includes:

[0045] S201: When a user device discovers a new application, the user device inputs the recommended information into a trained information recommendation model to obtain a probability of the user device clicking on the recommended information.

[0046] S202: Determine whether the information to be recommended is recommended in the application program based on the probability of the user device clicking on the information to be recommended.

[0047] In an embodiment of the present application, obtaining the probability of a user device clicking on the recommended information, i.e., predicting the click probability, means estimating the possible click probability of a certain information or advertisement before it is displayed in a certain situation, sorting the recommended information based on the click probability, and recommending the recommended information with a high probability to the user.

[0048] In an embodiment of the present application, when a new application is discovered, in order to enable the recommended information to be accurately recommended under the application, it is necessary to input the information to be recommended into the trained information recommendation model, and judge the probability of the user device clicking on the recommended information. If the information to be recommended is ranked high, the information to be recommended will be recommended in the application. If the information to be recommended is ranked low, the information to be recommended will not be recommended in the application. In this way, accurate recommendations can be made to users through the above steps.

[0049] By adopting the method for information recommendation provided by the present application, the probability of the user device clicking on the recommended information is obtained by inputting the information to be recommended into the trained information recommendation model, thereby realizing the feature characterization between the information to be recommended and the user device; based on the probability of the user device clicking on the recommended information predicted by the information recommendation model, it is determined whether the information to be recommended is recommended in the application, thereby realizing personalized recommendations for each individual without the need for excessive historical behavioral data, making information recommendations more accurate and providing a better user experience.

[0050] Optionally, combined Figure 3 As shown, according to the probability of the user device clicking on the recommended information, determining whether the recommended information is recommended in the application includes:

[0051] S301: The user device starts an application in response to a user's operation instruction;

[0052] S302: When the probability of the user device clicking on the recommended information is greater than a preset probability threshold, determine that the recommended information is recommended in the application.

[0053] In an embodiment of the present application, a probability threshold can be pre-set to determine whether the recommended information is recommended in the application. When recommending information to a user, the user device responds to the user's operation instruction, launches the application, and determines whether the probability of the user device clicking on the recommended information is greater than the preset probability threshold. When the calculated probability value is greater than the probability threshold, the recommended information is recommended in the application. When the calculated probability value is less than the probability threshold, the recommended information is not recommended in the application. Using the probability of the user device clicking on the recommended information obtained at this time, it is possible to accurately determine whether the user is interested in the recommended information.

[0054] Optionally, combined Figure 4 As shown, the information recommendation model is trained as follows:

[0055] S401, filtering feature information that meets set conditions from the application of the user device;

[0056] S402. The user device calculates a local model gradient of the information recommendation model based on the filtered feature information.

[0057] S403: The user device sends the local model gradient to the server;

[0058] S404: The user equipment receives the average model gradient sent by the server;

[0059] S405: The user equipment updates the information recommendation model using the average model gradient.

[0060] In an embodiment of the present application, feature information of applications in a user device is collected, feature information that meets set conditions is screened, and the feature information is input into a local information recommendation model to obtain a local model gradient. The information recommendation model includes an embedding layer, a first hidden layer, an interaction layer, a second hidden layer, and an output layer. In the process of obtaining the local model gradient, specifically, the feature information is input into the embedding layer to obtain the first feature output by the embedding layer, and then the first feature is input into the first hidden layer, and the output result is input into the interaction layer, and the second feature output by the interaction layer is input into the second hidden layer, and the output result is input into the output layer, thereby obtaining the final output result of the local information recommendation model. By repeating the above process, the local model gradient is obtained.

[0061] In this embodiment, the embedding layer maps discrete features into a dense vector space, such as mapping the feature "male" to [0.3, 0.4, ..., 0.9]. The first and second hidden layers form a nonlinear multilayer perceptron network. The interaction layer concatenates the user device's data and feature information, and the output layer predicts the probability of the user device clicking on the recommended information.

[0062] Optionally, combined Figure 5 As shown, filter the characteristic information that meets the set conditions from the application of this user's device

[0063] S501. Collect information about applications with recommended functions on the user's device;

[0064] S502: Based on the collected application information, filter feature information that meets the set conditions.

[0065] In the embodiment of the present application, in order to reduce the resources occupied by calculating the local model gradient, the application program with the recommendation function of the user device is determined, the application program with the recommendation function is combined, and the feature information of the above application programs is aggregated to screen the feature information that meets the set conditions.

[0066] Optionally, the feature information that is screened to meet the set conditions includes one or more of user attribute information, historical data information, and user interaction information.

[0067] In an embodiment of the present application, the user device is trained with one or more of user attribute information, historical data information, and user interaction information to obtain the probability of the user device clicking on the recommended information, and then obtain the local model gradient of the user device.

[0068] In the embodiments of the present application, user attribute information refers to the vectorized representation of personal attribute data in the user portrait and device attribute data in the user device. Personal attribute data includes the user's age data, gender data, marital data, occupation data, etc.; device attribute data includes the device's model data, storage data, operation data, system data, etc. These data are all discrete data, and the user's attribute information is obtained by vectorizing the discrete data. Historical data information refers to the historical record information of the user's device. The historical record information includes the user's text time series, video time series, such as comment behavior, etc. The characteristics of historical behavior are the vectorized representation of historical behavior data. Historical behavior data is the data generated by the comprehensive user's interactive behavior with respect to historical information.

[0069] In an embodiment of the present application, user interaction information refers to the characteristics of the user's historical behavior. User interaction information includes the user's interactive behavior with respect to historical information, such as purchasing behavior, clicking behavior, commenting behavior, forwarding behavior, liking behavior, etc. The information on historical behavior is a vectorized representation of historical behavior data. Historical behavior data is data generated by comprehensively analyzing the user's interactive behavior with respect to historical information.

[0070] Optionally, combined Figure 6 As shown, the local model gradient is sent to the server, including:

[0071] S601. Encrypt local model gradients through homomorphic encryption, differential privacy, or secret sharing.

[0072] S602: Send the encrypted local model gradient to the server.

[0073] In the embodiments of the present application, the leakage of model gradients or model parameters may lead to the leakage of important data and model information. Therefore, in order to ensure the privacy and data security of the user device and prevent the leakage of the data structure, the local model gradient is encrypted through homomorphic encryption, differential privacy, or secret sharing. Among them, the data is processed by the above encryption method to obtain an output, and this output is decrypted. The result is the same as the output result obtained by processing the unencrypted original data using the same method. In this way, the local model gradient can be sent to the server while protecting the privacy of the user.

[0074] Optionally, combined Figure 7 As shown, the average model gradient is used to update the information recommendation model, including:

[0075] S701, decrypting the average model gradient;

[0076] S702, updating the information recommendation model according to the decrypted average model gradient;

[0077] S703: until the loss value of the information recommendation model converges.

[0078] In the embodiments of the present application, since the above encryption method decrypts the data output, the result is the same as the output obtained by processing the unencrypted original data using the same method. Therefore, the decrypted average model gradient is obtained, and the parameters of the information recommendation model are updated based on this average model gradient. In this way, the model parameters can be adjusted according to the average model gradient until the model loss value converges, thereby obtaining the information recommendation model.

[0079] In an embodiment of the present application, in an embodiment of the present application, the cosine distance between user device information and user attribute information, historical data information or user interaction information is calculated; and the cosine distance is used as the loss value of the information recommendation model.

[0080] The larger the cosine distance, the more convergent the information recommendation model's loss value is, and the higher the probability that the user device will click on the recommended information. Based on this, the information recommendation model can be trained based on the loss value to obtain the user device's click rate on the recommended information.

[0081] Combine Figure 8 As shown, the embodiment of the present disclosure provides another method for information recommendation, which is applied to a server and includes:

[0082] S801. Receive n encrypted local model gradients sent by n user devices, where n is an integer greater than 1.

[0083] S802, calculating the average model gradient of n encrypted local model gradients;

[0084] S803: Send the average model gradient to each user device, so that the n user devices can obtain the probability of clicking on the information to be recommended, and determine whether the information to be recommended is recommended in the application.

[0085] In an embodiment of the present application, the server may use a parameter server or an aggregation server to perform aggregation calculation on n encrypted local model gradients. For example, the parameter server may calculate the weighted average of n encrypted local model gradients. Where n is an integer greater than 1; after calculating the average model gradient of the n encrypted local model gradients, the average model gradient is sent to each user device so that the n user devices can obtain the probability of clicking on the recommended information and determine whether the recommended information is recommended in the application.

[0086] Combine Figure 9 As shown, the embodiment of the present disclosure provides another method for information recommendation, including:

[0087] S901. Collect information about applications with recommended functions on the user device;

[0088] S902: Based on the collected application information, the user device filters feature information that meets the set conditions;

[0089] S903. The user equipment calculates a local model gradient of the information recommendation model based on the filtered feature information;

[0090] S904. The user device encrypts the local model gradient through homomorphic encryption, differential privacy, or secret sharing;

[0091] S905: The user device sends the encrypted local model gradient to the server.

[0092] S906. The server receives n encrypted local model gradients sent by n user devices;

[0093] S907. The server calculates the average model gradient of n encrypted local model gradients.

[0094] S908. The server sends the average model gradient to each user device;

[0095] S909: The user equipment receives the average model gradient sent by the server;

[0096] S9010. The user equipment decrypts the average model gradient.

[0097] S9011. The user equipment updates the information recommendation model according to the decrypted average model gradient until the loss value of the information recommendation model converges.

[0098] S9012. When the user device discovers a new application, the user device inputs the recommended information into the trained information recommendation model to obtain the probability of the user device clicking on the recommended information.

[0099] S9013. The user equipment starts the application in response to the user's operation instruction.

[0100] S9014: When the probability of the user device clicking on the information to be recommended is greater than a preset probability threshold, determine that the information to be recommended is recommended to the started application program.

[0101] In this way, by updating multiple user local information recommendation models, the most appropriate recommendations can be made to users when opening new applications; when new information to be recommended appears, the user device can also calculate whether to display it in the corresponding application based on the local information recommendation model.

[0102] Combine Figure 10 As shown, an embodiment of the present disclosure provides an apparatus for information recommendation, which is provided in a user device and includes:

[0103] The probability prediction module 1001 is configured to input the recommended information into the trained information recommendation model when a new application is discovered, and obtain the probability of the user device clicking on the recommended information;

[0104] The information recommendation module 1002 is configured to determine whether the information to be recommended is recommended in the application according to the probability of the user device clicking on the information to be recommended.

[0105] Combine Figure 11 As shown, the embodiment of the present disclosure provides another device for information recommendation, which is set on a server and includes:

[0106] a gradient receiving module configured to receive n encrypted local model gradients sent by n user devices; wherein n is an integer greater than 1;

[0107] A gradient calculation module is configured to calculate an average model gradient of n encrypted local model gradients;

[0108] The gradient sending module is configured to send the average model gradient to each user device, so that the n user devices can obtain the probability of clicking on the recommended information and determine whether the recommended information is recommended in the application.

[0109] By using the device for information recommendation provided by the present application, the probability of the user device clicking on the recommended information is obtained by inputting the information to be recommended into the trained information recommendation model, thereby realizing the feature characterization between the information to be recommended and the user device; based on the probability of the user device clicking on the recommended information predicted by the information recommendation model, it is determined whether the information to be recommended is recommended in the application, thereby realizing personalized recommendations for each individual, making information recommendations more accurate and providing a better user experience.

[0110] Combine Figure 12 As shown, an embodiment of the present disclosure provides a device for information recommendation, including a processor 100 and a memory 101. Optionally, the device may further include a communication interface 102 and a bus 103. The processor 100, the communication interface 102, and the memory 101 may communicate with each other via the bus 103. The communication interface 102 may be used for information transmission. The processor 100 may call the logic instructions in the memory 101 to execute the method for information recommendation of the above embodiment.

[0111] In addition, the logic instructions in the memory 101 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0112] Memory 101, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of the present disclosure. Processor 100 executes the program instructions / modules stored in memory 101 to perform functional applications and data processing, thereby implementing the information recommendation methods in the above-described embodiments.

[0113] The memory 101 may include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal device. Furthermore, the memory 101 may include high-speed random access memory and non-volatile memory.

[0114] An embodiment of the present disclosure provides an electronic device, comprising the above-mentioned device for information recommendation.

[0115] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned method for information recommendation.

[0116] An embodiment of the present disclosure provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions, which, when executed by a computer, cause the computer to execute the above-mentioned method for information recommendation.

[0117] The aforementioned computer-readable storage medium may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0118] The technical solution of the embodiments of the present disclosure may be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present disclosure. The aforementioned storage medium may be a non-transitory storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program code, or a transient storage medium.

[0119] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless explicitly required, individual components and functions are optional, and the order of operations may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to also include plural forms. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups of these. In the absence of further restrictions, an element defined by the sentence "comprising a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be found in the description of the method part.

[0120] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0121] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units can be merely a logical functional division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, and can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected to implement this embodiment according to actual needs. In addition, the functional units in the embodiments of the present disclosure may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0122] The flowcharts and block diagrams in the accompanying drawings show the possible implementation architectures, functions and operations of the systems, methods and computer program products according to the embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical functions. In some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified function or action, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A method for information recommendation, applied to a user device, characterized in that: include: When a new application is discovered, the recommended information is input into the trained information recommendation model to obtain the probability of the user device clicking on the recommended information; Determining whether the information to be recommended is recommended in the application based on the probability of the user device clicking on the information to be recommended; The information recommendation model is trained based on a federated learning algorithm, which integrates the data of various applications on the user's device and trains the information recommendation model in the following way: Filter characteristic information that meets the set conditions from the applications on the user's device; Calculate the local model gradient of the information recommendation model based on the filtered feature information; Send local model gradients to the server; Receive the average model gradient sent by the server; Use the average model gradient to update the information recommendation model; The feature information that meets the set conditions is filtered from the application of the user's device, including: Collect information about applications with recommended functions on the user's device; Based on the collected application information, filter the characteristic information that meets the set conditions; The feature information that is screened to meet the set conditions includes one or more of user attribute information and user interaction information.

2. The method according to claim 1, characterized in that Determining whether the recommended information is recommended in the application program based on a probability of the user device clicking on the recommended information includes: In response to a user's operation instruction, launching an application program; When the probability of the user equipment clicking on the recommended information is greater than a preset probability threshold, it is determined that the recommended information is recommended in the application.

3. The method according to claim 1, characterized in that The feature information that is screened to meet the set conditions also includes historical data information.

4. The method according to claim 1, wherein Send local model gradients to the server, including: Encrypt local model gradients via homomorphic encryption, differential privacy, or secret sharing; Send the encrypted local model gradients to the server.

5. A method for information recommendation, applied to a server, characterized in that: include: Receive n encrypted local model gradients sent by n user devices; where n is an integer greater than 1; Calculate the average model gradient of n encrypted local model gradients; Sending the average model gradient to each user device so that n user devices can obtain the probability of clicking on the information to be recommended, and determine whether the information to be recommended is recommended in the application; Each of the n user devices trains an information recommendation model based on a federated learning algorithm. The data from the various applications on the user devices is integrated, and the user devices calculate the local model gradient of the information recommendation model based on the filtered feature information. The feature information is obtained by filtering the applications of the user devices that meet the set conditions. Among them, filtering the applications on the user's device that meet the set conditions includes: Collect information about applications on user devices that have recommended functions; Based on the collected application information, filter the characteristic information that meets the set conditions; The feature information that is screened to meet the set conditions includes one or more of user attribute information and user interaction information.

6. A device for information recommendation, provided in a user device, characterized in that: include: A probability prediction module is configured to, when a new application is discovered, input the recommended information into a trained information recommendation model to obtain the probability of the user device clicking on the recommended information; The information recommendation module is configured to determine whether the recommended information is recommended in the application according to the probability of the user device clicking on the recommended information. The information recommendation model is trained based on a federated learning algorithm, which integrates the data of various applications on the user's device and trains the information recommendation model in the following way: Filter characteristic information that meets the set conditions from the applications on the user's device; Calculate the local model gradient of the information recommendation model based on the filtered feature information; Send local model gradients to the server; Receive the average model gradient sent by the server; Use the average model gradient to update the information recommendation model, The feature information that meets the set conditions is filtered from the application of the user's device, including: Collect information about applications with recommended functions on the user's device; Based on the collected application information, filter the characteristic information that meets the set conditions; The feature information that is screened to meet the set conditions includes one or more of user attribute information and user interaction information.

7. A device for information recommendation, provided on a server, characterized in that: include: a gradient receiving module configured to receive n encrypted local model gradients sent by n user devices; wherein n is an integer greater than 1; A gradient calculation module is configured to calculate an average model gradient of n encrypted local model gradients; a gradient sending module configured to send the average model gradient to each user device, so that the n user devices can obtain the probability of clicking on the to-be-recommended information and determine whether the to-be-recommended information is recommended in the application; Each of the n user devices trains an information recommendation model based on a federated learning algorithm. The data from the various applications on the user devices is integrated, and the user devices calculate the local model gradient of the information recommendation model based on the filtered feature information. The feature information is obtained by filtering the applications of the user devices that meet the set conditions. Among them, filtering the applications on the user's device that meet the set conditions includes: Collect information about applications on user devices that have recommended functions; Based on the collected application information, filter the characteristic information that meets the set conditions; The feature information that is screened to meet the set conditions includes one or more of user attribute information and user interaction information.

8. An electronic device comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the method for information recommendation according to any one of claims 1 to 5 when running the program instructions.

9. A storage medium storing program instructions, characterized in that: When the program instructions are executed, the method for information recommendation according to any one of claims 1 to 5 is executed.

Images