Multi-platform joint user recommendation method, device and electronic device

Abstract

The present invention provides a multi-platform joint user recommendation method, device and electronic device, which relates to the field of computer technology. The method is applied to a user terminal (including at least two social platforms) connected to a blockchain communication; the method includes: obtaining an initialized user recommendation model pre-encrypted and uploaded to the blockchain by a server; and performing initial training on the model based on user information to obtain at least two sets of first training parameters, and sending at least two first training parameters to the blockchain respectively; obtaining a second training parameter processed by the server, and updating the model after the initial training based on the second training parameter until the user recommendation model of each social platform reaches the preset model accuracy; and performing joint user recommendation on multiple platforms based on the user recommendation model that reaches the preset model accuracy. The present invention ensures the confidentiality and security of data between platforms, and effectively improves the user recommendation effect of the social platform.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a multi-platform joint user recommendation method, device and electronic device. Background Art

[0002] The emergence of Internet social platforms has brought great convenience to people's daily communication. Different social platforms have given users different social needs, so the same user often has his own account on different social platforms. At present, each social platform recommends users based on its own data. For some social platforms with relatively small amounts of data, it may be difficult to make more accurate user recommendations. In addition, since the data of different social platforms involve user privacy and commercial secrets, the user recommendation data between multiple platforms cannot be directly used by each other. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-platform joint user recommendation method, device and electronic device, which can train a user recommendation model without exchanging user information of each platform, thereby ensuring the confidentiality and security of data between platforms. For social platforms with less data, the multi-platform joint user recommendation method can be used to effectively improve the user recommendation effect of the social platform.

[0004] In a first aspect, the present invention provides a multi-platform joint user recommendation method, which is applied to a user terminal connected to a blockchain communication; the user terminal includes at least two social platforms; the method includes: obtaining an initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by a server; performing initial training on the initialized user recommendation model based on user information to obtain at least two sets of first training parameters of the user recommendation model after the initial training, and sending the at least two sets of first training parameters to the blockchain respectively; obtaining second training parameters processed by the server, and updating the user recommendation model after the initial training based on the second training parameters until the user recommendation model of each social platform reaches the preset model accuracy; the second training parameters are obtained based on at least two sets of first training parameters; and performing multi-platform joint user recommendation based on the user recommendation model that reaches the preset model accuracy.

[0005] In an optional embodiment, before obtaining the initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by the server, the method also includes: constructing a user network based on platform text data; the user network is a network with user nodes of the social platform and relationships between users as edges; performing preset matrix processing on the user network of each social platform to obtain user information with the same dimension corresponding to each social platform; the user information includes a user name and user feature information corresponding to the user name.

[0006] In an optional embodiment, the step of obtaining an initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by a server includes: receiving a first public key sent by the server; each social platform corresponds to a first public key; based on the first public key, the initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by the server is decrypted, and downloaded to the local storage of each social platform respectively; wherein the initialized user recommendation model is encrypted by the server using a first private key corresponding to the first public key and uploaded to the blockchain.

[0007] In an optional embodiment, the step of performing initial training on an initialized user recommendation model based on user information to obtain first training parameters of the user recommendation model after initial training, and sending the first training parameters to a blockchain includes: for each social platform, performing initial training on the initialized user recommendation model downloaded to local storage of the social platform based on user information corresponding to the social platform to obtain first training parameters; the initialized user recommendation model of each social platform corresponds to the first training parameters obtained after initial training; and encrypting at least two groups of first training parameters obtained after initial training of at least two social platforms using a second private key, and sending the encrypted at least two groups of first training parameters to the blockchain.

[0008] In an optional embodiment, the second training parameter is encrypted by the third private key of the server and uploaded to the blockchain; the second training parameter processed by the server is obtained, and the user recommendation model after initial training is updated and trained based on the second training parameter until the user recommendation model of each social platform reaches the preset model accuracy, including: for each social platform, the second training parameter processed by the server is obtained based on the third public key corresponding to the third private key; wherein each social platform corresponds to a third public key; the second training parameter is obtained by performing a mean aggregation operation based on at least two groups of first training parameters; the user recommendation model after initial training is updated and trained based on the second training parameter until the user recommendation model of each social platform reaches the preset model accuracy.

[0009] In a second aspect, the present invention provides a multi-platform joint user recommendation method, which is applied to a server connected to a blockchain; the method includes: encrypting an initialized user recommendation model through a first private key and uploading it to the blockchain, and sending a first public key corresponding to the first private key to at least two social platforms, so that the at least two social platforms obtain the initialized user recommendation model by decrypting it through the first public key; obtaining at least two groups of first training parameters after initial training of the initialized user recommendation model by at least two social platforms from the blockchain based on the second public key; the first training parameters are encrypted and uploaded to the blockchain by at least two second private keys corresponding to the second public key; performing a mean aggregation operation on at least two groups of first training parameters to obtain second training parameters; encrypting the second training parameters based on the third private key and uploading them to the blockchain, and sending the third public key corresponding to the third private key to at least two social platforms.

[0010] In a third aspect, the present invention provides a multi-platform joint user recommendation device, the device is arranged on a user terminal connected to a blockchain communication; the user terminal includes at least two social platforms; the device includes: an acquisition module, used to obtain an initialized user recommendation model pre-encrypted and uploaded to the blockchain by a server; an initial training module, used to perform initial training on the initialized user recommendation model based on user information, obtain at least two sets of first training parameters of the user recommendation model after the initial training, and send the at least two sets of first training parameters to the blockchain respectively; an update training module, used to obtain second training parameters processed by the server, and update the user recommendation model after the initial training based on the second training parameters until the user recommendation model of each social platform reaches the preset model accuracy; the second training parameters are obtained based on at least two sets of first training parameters; a joint recommendation module, used to perform multi-platform joint user recommendation based on the user recommendation model that reaches the preset model accuracy.

[0011] In a fourth aspect, the present invention provides a multi-platform joint user recommendation device, which is arranged on a server connected to a blockchain; the device includes: an initial encryption upload module, which is used to encrypt and upload the initialized user recommendation model to the blockchain through a first private key, and send the first public key corresponding to the first private key to at least two social platforms, so that the at least two social platforms can decrypt the initialized user recommendation model through the first public key; a parameter acquisition module, which is used to obtain at least two groups of first training parameters of the initialized user recommendation model after initial training by at least two social platforms from the blockchain based on the second public key; the first training parameters are encrypted and uploaded to the blockchain by at least two second private keys corresponding to the second public key; a parameter processing module, which is used to perform a mean aggregation operation on at least two groups of first training parameters to obtain second training parameters; an update encryption upload module, which is used to encrypt and upload the second training parameters to the blockchain based on a third private key, and send the third public key corresponding to the third private key to at least two social platforms.

[0012] In a fifth aspect, the present invention provides an electronic device, comprising a processor and a memory; a computer program is stored in the memory, and when the computer program is run by the processor, a multi-platform joint user recommendation method as described in any one of the aforementioned implementations is executed.

[0013] In a sixth aspect, the present invention provides a computer-readable storage medium for storing computer software instructions used in the multi-platform joint user recommendation method as described in any of the aforementioned implementations.

[0014] The multi-platform joint user recommendation method, device and electronic device provided by the present invention, for a method applied to a user terminal (including at least two social platforms) connected to a blockchain communication, first obtains an initialized user recommendation model pre-encrypted and uploaded to the blockchain by a server, and then performs initial training on the initialized user recommendation model based on user information, obtains at least two (i.e., one-to-one corresponding to the number of social platforms) first training parameters of the user recommendation model after initial training, and sends at least two groups of first training parameters to the blockchain respectively, obtains the second training parameters processed by the server, wherein the second training parameters are obtained based on at least two groups of first training parameters, and then updates the user recommendation model after initial training based on the second training parameters, until the user recommendation model of each social platform reaches the preset model accuracy, so as to finally perform multi-platform joint user recommendation based on the user recommendation model that reaches the preset model accuracy. In the above method, in the process of data interaction between the user terminal and the server, the blockchain is used as an intermediate medium, and the user recommendation model can be trained without exchanging user information of each platform, thereby ensuring the confidentiality and security of data between platforms. For social platforms with less data, the user recommendation effect of the social platform can be effectively improved by the multi-platform joint user recommendation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A schematic diagram of a flow chart of a multi-platform joint user recommendation method provided by an embodiment of the present invention;

[0017] Figure 2 A flowchart of another multi-platform joint user recommendation method provided by an embodiment of the present invention;

[0018] Figure 3 A flowchart of a specific multi-platform joint user recommendation method provided by an embodiment of the present invention;

[0019] Figure 4 A schematic diagram of the structure of a multi-platform user recommendation device provided by an embodiment of the present invention;

[0020] Figure 5 A schematic diagram of the structure of another multi-platform user recommendation device provided by an embodiment of the present invention;

[0021] Figure 6 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0023] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0025] In the description of the present invention, it should be noted that the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0026] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0027] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0028] Considering that each social platform currently makes user recommendations based on its own data, it may be difficult to make more accurate user recommendations for some social platforms with relatively small amounts of data. In addition, since the data of different social platforms involve user privacy and commercial secrets, the user recommendation data between multiple platforms cannot be directly used by each other. The embodiments of the present invention provide a multi-platform joint user recommendation method, device and electronic device, which can train a user recommendation model without exchanging user information of each platform, thereby ensuring the confidentiality and security of data between platforms. For social platforms with less data, the multi-platform joint user recommendation method can be used to effectively improve the user recommendation effect of the social platform.

[0029] For ease of understanding, a multi-platform joint user recommendation method provided by an embodiment of the present invention is first described in detail. Figure 1 The flowchart of a multi-platform joint user recommendation method is shown in FIG. 1 , which is applied to a user terminal connected to a blockchain communication, and the user terminal includes at least two social platforms, and mainly includes the following steps S102 to S108:

[0030] Step S102, obtaining an initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by the server.

[0031] The server is connected to the blockchain for communication. In this embodiment, the data interaction between the server and the social platform of the user terminal is completed through the blockchain. The initialized user recommendation model uploaded to the blockchain is encrypted by the server using a private key, and the public key corresponding to the private key is distributed to at least two social platforms accordingly. The number of social platforms can usually be determined according to actual conditions. Since joint user recommendations between multiple platforms are required, there are at least two social platforms.

[0032] Step S104: Initially train the initialized user recommendation model based on the user information to obtain at least two sets of first training parameters of the user recommendation model after the initial training, and send the at least two sets of first training parameters to the blockchain respectively.

[0033] It is understandable that a social platform usually includes a user name and user feature information corresponding to the user name, and the user feature information may be information such as the user's hobbies, features, etc. The user name of each user is set corresponding to the user feature information so that user recommendations can be made based on the same user feature information (such as hobbies). In this embodiment, when at least two social platforms train the initialized user recommendation model, they may use training samples of the same dimension, such as when the user information (i.e., the user name and the user feature information corresponding to the user name) is represented in the form of a matrix, the matrix format of the user information of each social platform is the same.

[0034] When the social platform obtains the initialized user recommendation model, the model can be trained based on user information so that the user recommendation model can meet the user recommendation needs of the current social platform. In this embodiment, the initialized user recommendation model obtained from the blockchain by at least two social platforms is the same user recommendation model, and the initialized user recommendation model is trained by different social platforms to obtain at least two groups of first training parameters, wherein each social platform corresponds to a group of first training parameters, and the number of groups of first training parameters sent to the blockchain corresponds to the number of social platforms.

[0035] Step S106, obtaining the second training parameters processed by the server, and updating the user recommendation model after the initial training based on the second training parameters until the user recommendation model of each social platform reaches the preset model accuracy.

[0036] After each social platform sends the first training parameters obtained through training to the blockchain, the server processes the first training parameters of multiple social platforms, that is, the second training parameters are obtained based on at least two groups of first training parameters. In one embodiment, in order to make the training parameters applicable to any social platform, the server may perform mean aggregation on at least two groups of first training parameters to obtain second training parameters, the number of which is one group, rather than at least two groups of first training parameters, and the training parameters of this group are applicable to the user recommendation model of any social platform.

[0037] It is understandable that usually one training (i.e. obtaining the first training parameters) and one mean aggregation cannot achieve the best effect of the model. Therefore, the training process can be performed multiple times until the user recommendation model of each social platform reaches the preset model accuracy.

[0038] Step S108, performing multi-platform joint user recommendation based on the user recommendation model that reaches the preset accuracy of the model.

[0039] Joint user recommendation is the process of performing cross-platform user recommendation using the user recommendation model with the same second training parameters obtained through the above training process. After determining that the user recommendation model of each social platform reaches the preset model accuracy, the user recommendation model is used to perform multi-platform joint user recommendation.

[0040] The multi-platform joint user recommendation method provided in the embodiment of the present invention, when training the user recommendation model, in the process of data interaction between the user terminal and the server, uses the blockchain as an intermediate medium, and can train the user recommendation model without exchanging user information of each platform, thereby ensuring the confidentiality and security of data between platforms. For social platforms with less data, the multi-platform joint user recommendation method can be used to effectively improve the user recommendation effect of the social platform.

[0041] In order to ensure the stability and accuracy of the training process for at least two social platforms, before obtaining the initialized user recommendation model pre-encrypted and uploaded to the blockchain by the server, it is necessary to process the user information of each platform. In one embodiment, the user information includes a user name and user feature information corresponding to the user name, and the user information is represented by a matrix. Then, the processing process is to process the user information of at least two social platforms into matrix information with the same dimension. For example, a user network can be first constructed based on the platform text data. The platform text data is the user name and user features recorded in text on the social platform. The user name can be the user's name, nickname, etc. The user feature information can include the user's hobbies, age, things of interest, etc. The user network is a network with user nodes and relationships between users of the social platform as edges. The main relationship diagram between users can include friends, attention, likes, forwarding, etc. If the matrix dimensions of the user information of each social platform are different, the user network of each social platform is processed by a preset matrix processing method to obtain user information with the same dimension corresponding to each social platform. The method of processing the matrix into the same dimension can be determined according to the actual situation.

[0042] In one embodiment, when a social platform obtains an initialized user recommendation model that is pre-encrypted and uploaded to a blockchain by a server, it may first receive a first public key sent by the server. It is understandable that, in order to ensure that important data of the platform is not leaked, each social platform corresponds to a first public key. Then, the social platform decrypts the initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by the server based on the first public key, and downloads it to the local storage of each social platform respectively, wherein the initialized user recommendation model is encrypted by the server using a first private key corresponding to the first public key and uploaded to the blockchain.

[0043] When each social platform receives the initialized user recommendation model, for each social platform, based on the user information corresponding to the social platform, which is also the user information of the same dimension between the above-mentioned social platforms, each social platform then uses its own user information to perform initial training on the initialized user recommendation model downloaded to the local storage of the social platform to obtain the first training parameters. It can be understood that the initialized user recommendation model of each social platform corresponds to the first training parameters obtained after the initial training, and the first training parameters obtained after the training of each social platform are not the same. Similarly, in order to ensure the reliability of the data, at least two groups of first training parameters obtained after the initial training of at least two social platforms can be encrypted by the second private key, and the second public key corresponding to the second private key can be sent to the server, and the encrypted at least two groups of first training parameters can be sent to the blockchain.

[0044] In one embodiment, the server decrypts the encrypted first training parameters uploaded to the blockchain by using the received second public key, and further processes (such as the above-mentioned mean aggregation process) to obtain the second training parameters, and then encrypts the second training parameters by using the third private key through the server and uploads them to the blockchain, and sends the third public key corresponding to the third private key to at least two social platforms, wherein each social platform corresponds to a third public key. For each social platform, the second training parameters processed by the server are obtained based on the third public key corresponding to the third private key, and then the user recommendation model after the initial training is updated and trained based on the obtained second training parameters, until the user recommendation model of each social platform reaches the preset model accuracy, so as to perform user recommendations on multiple platforms through the user recommendation model that reaches the preset model accuracy.

[0045] The embodiment of the present invention also provides another multi-platform joint user recommendation method, which is applied to a server connected to the blockchain communication, see Figure 2 The flowchart of another multi-platform joint user recommendation method shown in FIG. 1 mainly includes the following steps S202 to S208:

[0046] Step S202: encrypt the initialized user recommendation model with the first private key and upload it to the blockchain, and send the first public key corresponding to the first private key to at least two social platforms, so that the at least two social platforms can decrypt the initialized user recommendation model with the first public key.

[0047] Step S204: obtaining at least two sets of first training parameters of at least two social platforms after initial training of the initialized user recommendation model from the blockchain based on the second public key; the first training parameters are encrypted and uploaded to the blockchain using at least two second private keys corresponding to the second public key.

[0048] Step S206: performing a mean aggregation operation on at least two groups of first training parameters to obtain second training parameters.

[0049] Step S208: encrypt the second training parameter based on the third private key and upload it to the blockchain, and send the third public key corresponding to the third private key to at least two social platforms.

[0050] The multi-platform joint user recommendation method provided in an embodiment of the present invention performs a mean aggregation operation on at least two groups of first training parameters by a server to obtain second training parameters, which can aggregate the training parameters of multiple social platforms so that the trained second training parameters are suitable for each social platform, thereby allowing at least two user recommendation models corresponding to the second parameters to be combined to achieve multi-platform joint user recommendation.

[0051] The method provided in the embodiment of the present invention, its implementation principle and the technical effect produced are similar to the above method ( Figure 1 For the sake of brief description, for matters not mentioned in the embodiment part, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0052] In addition, the embodiment of the present invention also provides a specific multi-platform joint user recommendation method, see Figure 3 The flowchart of a specific multi-platform joint user recommendation method is shown, and the method is interactively executed by a user terminal and a server, and mainly includes the following steps 302 to S314:

[0053] Step S302: After the server initializes the model, it encrypts it with a private key and uploads it to the blockchain, and sends the public key to the platforms participating in the training. First, the server encrypts the training model (i.e., the user recommendation model) and the initialized parameters with a private key and uploads them to the blockchain, and sends the public key to each social platform.

[0054] Step S304: Each platform builds a user network based on its own data, pre-processes the data of each platform, and downloads the model to the local computer through the public key to start training. The constructed user network refers to a network with platform users as nodes and the relationships between users in the platform as edges (user relationships on different platforms may be different, such as friends, followers, likes, reposts, etc.). By pre-processing the data of each platform, the data of at least two social platforms can have the same dimension.

[0055] In one embodiment, the preprocessing process of data from different social platforms can first uniformly set the network walk length l and the walk number t, and then obtain the walk sequence S(n) of all nodes in the user network according to the set walk length and walk number, and then calculate the walk sequence S(n) according to the formula Calculate the embedding vector of each node to obtain the embedding matrix f. Among them, V is the set of nodes in the user network, n is the initial node where the walk starts, and f(n) is the embedding vector of node n. Then integrate the obtained matrix f so that the final embedding matrices f of each platform are of the same size. Finally, the obtained embedding matrix is ​​used as the new adjacency matrix A of the user network, and the corresponding feature matrix X is obtained based on the new adjacency matrix A. Among them, the above-mentioned embedding matrix of the same size can first remove duplicate vectors from the matrix, and calculate the similarity of all remaining vectors based on cosine similarity, and merge vectors with higher similarity. During the merging process, align the embedding matrices of multiple platforms so that vectors with higher similarity are in the same row.

[0056] In one implementation, the user recommendation model can be expressed as follows by formula (1):

[0057]

[0058] Among them, A is the adjacency matrix used to identify user information, and Z is an N*F matrix, which can be obtained by Z=softmax(AReLU(AXW 0 )W 1 ) represents, where N is the number of nodes, F is the user category, W0 and W1 are the model parameters of the user recommendation model, such as, when the user recommendation model is a two-layer neural network model, W0 and W1 are the parameters of each layer of the network respectively; X is the feature matrix, the rows of the feature matrix are users, and the columns are different user features. User features can be some hobbies and features displayed by users on social platforms, such as QQ tags.

[0059] Step S306, each platform trains the model once using local data, encrypts the trained parameters with a private key and uploads them to the blockchain, and sends the public key to the server. The trained parameters are uploaded to the blockchain using private key encryption, and the public key corresponding to the private key is sent to the server so that the server can decrypt and obtain the trained parameters based on the public key. Local data refers to the preprocessed adjacency matrix A and node feature matrix X in the above formula. The parameters encrypted after training are W0 and W1 in the formula. Encryption methods such as homomorphic encryption can be used. In actual applications, they can also be selected according to actual needs, and are not specifically limited here.

[0060] Step S308, the server obtains the parameters of each platform through the public key, performs mean aggregation, encrypts the parameters with the private key, uploads them to the blockchain, and sends the corresponding public key to the platforms participating in the training. Mean aggregation refers to the operation of averaging the elements in the W0 and W1 matrices after each platform is trained once.

[0061] In step S310, each platform uses the public key to retrieve the updated parameters from the blockchain to update the model locally. The social platform decrypts the parameters using the public key to obtain the parameters after mean aggregation processing, and uses the updated parameters (i.e. the parameters after mean aggregation processing) to update the model locally.

[0062] Step S312, determine whether each client model reaches the preset accuracy, if yes, execute step S314, if no, execute step S306. Determine whether each platform reaches the preset training accuracy, if yes, stop training each platform for user recommendation, if not, each platform performs local training again until the preset accuracy is reached.

[0063] Step S314, the training ends, and each platform makes user recommendations based on the reconstructed adjacency matrix information.

[0064] The multi-platform joint user recommendation method provided by the embodiment of the present invention can perform user recommendation algorithm training without exchanging user data of each platform through the above steps S302 to S314, which can improve the recommendation effect of the social platform, and the effect is more obvious for social platforms with less data. At the same time, since both the user device and the server interact with the blockchain, data upload and data acquisition are performed by private key encryption and public key decryption, which further ensures the reliability and security of the data during the training process.

[0065] The present invention provides a multi-platform joint user recommendation device, see Figure 4 The schematic diagram of the structure of a multi-platform joint user recommendation device is shown, the device is set in a user terminal connected to the blockchain communication; the user terminal includes at least two social platforms; the device includes the following parts:

[0066] An acquisition module 402 is used to acquire an initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by a server;

[0067] An initial training module 404 is used to perform initial training on the initialized user recommendation model based on user information, obtain at least two sets of first training parameters of the user recommendation model after initial training, and send the at least two sets of first training parameters to the blockchain respectively;

[0068] An update training module 406 is used to obtain a second training parameter processed by the server, and update the user recommendation model after the initial training based on the second training parameter until the user recommendation model of each social platform reaches a preset model accuracy; the second training parameter is obtained based on at least two sets of first training parameters;

[0069] The joint recommendation module 408 is used to perform joint user recommendation on multiple platforms based on a user recommendation model that reaches a preset accuracy of the model.

[0070] The multi-platform joint user recommendation device provided by the embodiment of the present invention, when training the user recommendation model, uses blockchain as an intermediate medium in the process of data interaction between the user terminal and the server, and can train the user recommendation model without exchanging user information of each platform, thereby ensuring the confidentiality and security of data between platforms. For social platforms with less data, the multi-platform joint user recommendation method can be used to effectively improve the user recommendation effect of the social platform.

[0071] In one embodiment, before obtaining the initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by the server, the device also includes a user information processing module for constructing a user network based on platform text data; the user network is a network with user nodes of the social platform and the relationship between users as edges; the user network of each social platform is subjected to preset matrix processing to obtain user information with the same dimension corresponding to each social platform; the user information includes a user name and user feature information corresponding to the user name.

[0072] In one embodiment, the acquisition module 402 is also used to receive a first public key sent by the server; each social platform corresponds to a first public key; based on the first public key, the initialized user recommendation model pre-encrypted and uploaded to the blockchain by the server is decrypted, and downloaded to the local storage of each social platform respectively; wherein the initialized user recommendation model is encrypted by the server using the first private key corresponding to the first public key and then uploaded to the blockchain.

[0073] In one embodiment, the initial training module 404 is also used to perform initial training on the initialized user recommendation model downloaded to the local storage of the social platform for each social platform based on the user information corresponding to the social platform to obtain the first training parameters; the initialized user recommendation model of each social platform corresponds to the first training parameters obtained after the initial training; at least two groups of first training parameters obtained after the initial training of at least two social platforms are encrypted by the second private key, and the encrypted at least two groups of first training parameters are sent to the blockchain.

[0074] In one embodiment, the second training parameter is encrypted by the third private key of the server and uploaded to the blockchain; the update training module 406 is also used to obtain the second training parameter processed by the server for each social platform based on the third public key corresponding to the third private key; wherein each social platform corresponds to a third public key; the second training parameter is obtained by performing a mean aggregation operation based on at least two first training parameters; based on the second training parameter, the user recommendation model after initial training is updated and trained until the user recommendation model of each social platform reaches the preset model accuracy.

[0075] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as the aforementioned Figure 1 The method embodiments described are the same. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiments.

[0076] The embodiment of the present invention also provides another multi-platform joint user recommendation device, see Figure 5The schematic diagram of another multi-platform joint user recommendation device is shown, and the device is set on a server connected to the blockchain communication; the device includes the following parts:

[0077] An initial encryption upload module 502 is used to encrypt and upload the initialized user recommendation model to the blockchain using a first private key, and send a first public key corresponding to the first private key to at least two social platforms, so that the at least two social platforms can decrypt the initialized user recommendation model using the first public key;

[0078] The parameter acquisition module 504 is used to obtain at least two sets of first training parameters of the initialized user recommendation model after initial training of at least two social platforms from the blockchain based on the second public key; the first training parameters are encrypted and uploaded to the blockchain by at least two second private keys corresponding to the second public key;

[0079] A parameter processing module 506, configured to perform a mean aggregation operation on at least two groups of first training parameters to obtain second training parameters;

[0080] The updated encryption upload module 508 is used to encrypt the second training parameter based on the third private key and upload it to the blockchain, and send the third public key corresponding to the third private key to at least two social platforms.

[0081] The multi-platform joint user recommendation device provided in the embodiment of the present invention performs a mean aggregation operation on at least two groups of first training parameters through a server to obtain second training parameters, so that the training parameters of multiple social platforms can be aggregated so that the trained second training parameters are suitable for each social platform, thereby enabling at least two user recommendation models corresponding to the second parameters to be combined to achieve multi-platform joint user recommendation.

[0082] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as the aforementioned Figure 2 The method embodiments described are the same. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiments.

[0083] An embodiment of the present invention provides an electronic device, which may be a user terminal or a server. Specifically, the electronic device includes a processor and a storage device; a computer program is stored on the storage device, and when the computer program is run by the processor, it executes the method described in any one of the above-mentioned implementation methods.

[0084] Figure 6A structural diagram of an electronic device provided in an embodiment of the present invention, the electronic device 100 includes: a processor 60, a memory 61, a bus 62 and a communication interface 63, wherein the processor 60, the communication interface 63 and the memory 61 are connected via the bus 62; the processor 60 is used to execute an executable module stored in the memory 61, such as a computer program.

[0085] The memory 61 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 63 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0086] The bus 62 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0087] Among them, the memory 61 is used to store programs, and the processor 60 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 60 or implemented by the processor 60.

[0088] The processor 60 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 60. The above processor 60 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 61, and the processor 60 reads the information in the memory 61 and completes the steps of the above method in combination with its hardware.

[0089] The multi-platform joint user recommendation method, device and computer program product of the electronic device provided in the embodiments of the present invention include a computer-readable storage medium storing non-volatile program code executable by a processor, and a computer program is stored on the computer-readable storage medium. The computer program is executed by the processor to execute the method described in the previous method embodiments. The specific implementation can be found in the method embodiments, which will not be repeated here.

[0090] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned embodiment, and will not be repeated here.

[0091] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the method described in the previous method embodiment. The specific implementation can be referred to the method embodiment, which will not be repeated here.

[0092] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as 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.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-platform joint user recommendation method, characterized in that: The method is applied to a user terminal connected to a blockchain; the user terminal includes at least two social platforms; the method includes: Obtain the initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by the server; Performing initial training on the initialized user recommendation model based on the user information to obtain at least two sets of first training parameters of the user recommendation model after initial training, and sending at least two of the first training parameters to the blockchain respectively; Obtaining a second training parameter processed by the server, and updating and training the user recommendation model after the initial training based on the second training parameter until the user recommendation model of each social platform reaches a preset model accuracy; the second training parameter is obtained based on at least two of the first training parameters; Based on the user recommendation model that reaches the preset accuracy of the model, multi-platform joint user recommendation is performed.

2. The multi-platform user recommendation method according to claim 1, characterized in that: Before obtaining the initialized user recommendation model pre-encrypted and uploaded to the blockchain by the server, the method further includes: Constructing a user network based on platform text data; the user network is a network with user nodes of the social platform and relationships between users as edges; A preset matrix processing is performed on the user network of each social platform to obtain the user information having the same dimension as that corresponding to each social platform; the user information includes a user name and user feature information corresponding to the user name.

3. The multi-platform user recommendation method according to claim 1, characterized in that: The step of obtaining the initialized user recommendation model pre-encrypted and uploaded to the blockchain by the server includes: Receive a first public key sent by the server; each of the social platforms corresponds to one first public key; Based on the first public key, the initialized user recommendation model that was pre-encrypted and uploaded to the blockchain by the server is decrypted and downloaded to the local storage of each social platform respectively; wherein the initialized user recommendation model is encrypted by the server using the first private key corresponding to the first public key and then uploaded to the blockchain.

4. The multi-platform user recommendation method according to claim 3, characterized in that: The step of initially training the initialized user recommendation model based on the user information, obtaining a first training parameter of the user recommendation model after the initial training, and sending the first training parameter to the blockchain includes: For each of the social platforms, based on the user information corresponding to the social platform, the initialized user recommendation model downloaded to the local storage of the social platform is initially trained to obtain the first training parameters; the initialized user recommendation model of each of the social platforms corresponds to the first training parameters obtained after the initial training; The at least two groups of first training parameters obtained after the initial training of at least two of the social platforms are encrypted using a second private key, and the encrypted at least two groups of first training parameters are sent to the blockchain.

5. The multi-platform user recommendation method according to claim 1, characterized in that: The second training parameter is encrypted by the third private key of the server and uploaded to the blockchain; the step of obtaining the second training parameter processed by the server, and updating the user recommendation model after the initial training based on the second training parameter until the user recommendation model of each social platform reaches the preset model accuracy, includes: For each of the social platforms, obtaining a second training parameter processed by the server based on a third public key corresponding to the third private key; wherein each of the social platforms corresponds to a third public key; and the second training parameter is obtained by performing a mean aggregation operation based on at least two of the first training parameters; The user recommendation model after the initial training is updated and trained based on the second training parameters until the user recommendation model of each social platform reaches a preset model accuracy.

6. A multi-platform joint user recommendation method, characterized in that: The method is applied to a server connected to a blockchain communication; the method comprises: Encrypting the initialized user recommendation model with a first private key and uploading it to the blockchain, and sending a first public key corresponding to the first private key to at least two social platforms, so that at least two of the social platforms can decrypt the initialized user recommendation model with the first public key; Based on the second public key, at least two groups of first training parameters after initial training of the initialized user recommendation model by at least two social platforms are obtained from the blockchain; the first training parameters are encrypted and uploaded to the blockchain by at least two second private keys corresponding to the second public key; Performing a mean aggregation operation on at least two groups of first training parameters to obtain second training parameters; The second training parameter is encrypted based on a third private key and uploaded to the blockchain, and a third public key corresponding to the third private key is sent to the at least two social platforms.

7. A multi-platform user recommendation device, characterized in that: The device is arranged in a user terminal connected to the blockchain communication; the user terminal includes at least two social platforms; the device includes: An acquisition module, used to acquire an initialized user recommendation model that is pre-encrypted and uploaded to the blockchain by a server; An initial training module, used to perform initial training on the initialized user recommendation model based on user information, obtain at least two sets of first training parameters of the user recommendation model after initial training, and send at least two of the first training parameters to the blockchain respectively; An update training module, used to obtain a second training parameter processed by the server, and update the user recommendation model after the initial training based on the second training parameter until the user recommendation model of each social platform reaches a preset model accuracy; the second training parameter is obtained based on at least two of the first training parameters; The joint recommendation module is used to perform joint user recommendation on multiple platforms based on the user recommendation model that reaches the preset accuracy of the model.

8. A multi-platform user recommendation device, characterized in that: The device is arranged on a server connected to the blockchain communication; the device comprises: An initial encryption upload module, used to encrypt and upload the initialized user recommendation model to the blockchain using a first private key, and send a first public key corresponding to the first private key to at least two social platforms, so that at least two of the social platforms can decrypt the initialized user recommendation model using the first public key; A parameter acquisition module, configured to acquire, from the blockchain based on a second public key, at least two sets of first training parameters after initial training of the initialized user recommendation model by at least two social platforms; the first training parameters are encrypted and uploaded to the blockchain using at least two second private keys corresponding to the second public key; A parameter processing module, used for performing a mean aggregation operation on at least two groups of first training parameters to obtain second training parameters; An updated encryption upload module is used to encrypt the second training parameter based on a third private key and upload it to the blockchain, and send a third public key corresponding to the third private key to the at least two social platforms.

9. An electronic device, characterized in that: including a processor and a memory; The memory stores a computer program, which, when executed by the processor, executes the multi-platform joint user recommendation method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: Used to store computer software instructions used in the multi-platform user recommendation method according to any one of claims 1 to 6.

Images