CN115718930A - User service system and method based on user portrait and electronic equipment - Google Patents
User service system and method based on user portrait and electronic equipment Download PDFInfo
- Publication number
- CN115718930A CN115718930A CN202211526982.9A CN202211526982A CN115718930A CN 115718930 A CN115718930 A CN 115718930A CN 202211526982 A CN202211526982 A CN 202211526982A CN 115718930 A CN115718930 A CN 115718930A
- Authority
- CN
- China
- Prior art keywords
- user
- portrait
- model
- user portrait
- network parameters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012549 training Methods 0.000 claims abstract description 62
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 14
- 238000000586 desensitisation Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 230000002776 aggregation Effects 0.000 claims description 6
- 238000004220 aggregation Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 abstract description 12
- 230000003993 interaction Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000009434 installation Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Bioethics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application provides a user service system and method based on user portrait and electronic equipment. The user service system based on the user portrait comprises a plurality of clients and a central server; each client is used for determining the intermediate local network parameters of the updated initial user portrait local model; training the updated initial user portrait local model again to determine a trained target user portrait local model; a central server; and determining a target user portrait global model based on each target local network parameter sent by each client and the updated initial user portrait global model so as to use the target user portrait global model to recommend services to each target user. According to the method and the device, the network parameters of the models in the clients are uploaded instead of uploading target user data in the training process of the initial user portrait local model, so that the purpose of privacy protection of the client data is achieved, and the data security is improved.
Description
Technical Field
The present application relates to the field of data processing technologies, and in particular, to a user service system and method based on a user portrait, and an electronic device.
Background
With the development of social science and technology and the improvement of the living standard of people, more and more people start to use various clients, and therefore, the data security and data privacy disclosure problems of the clients in the circulation process face a lot of challenges.
The traditional copy-type data circulation mode seriously leaks business privacy information, personal privacy information and the like, and cannot meet the requirements of privacy protection and data security of corresponding data of a user.
Disclosure of Invention
In view of the above, an object of the present application is to provide a user service system, a method and an electronic device based on a user portrait, in which target user data in a process of training a local model of an initial user portrait is not uploaded, but network parameters of models in clients are uploaded, so as to achieve the purpose of protecting privacy of client data, thereby improving security of data.
In a first aspect, an embodiment of the present application provides a user service system based on a user representation, which includes a plurality of clients and a central server:
each client is used for training the constructed initial user portrait local model based on the acquired target user data, determining intermediate local network parameters of the updated initial user portrait local model, and sending the intermediate local network parameters to the central server; training the updated initial user portrait local model again according to the intermediate global network parameters sent by the central server until the trained target user portrait local model is determined, and sending target local network parameters corresponding to the target user portrait local model to the central server;
the central server is used for training the constructed initial user portrait global model according to the plurality of intermediate local network parameters sent by the plurality of clients, determining the intermediate global network parameters of the updated initial user portrait global model, and sending the intermediate global network parameters to the clients; and determining a target user portrait global model based on each target local network parameter sent by each client and the updated initial user portrait global model, so that each client can divide user types and recommend services for each target user by using the target user portrait global model.
Further, the client determines the target user data by:
acquiring initial user data;
and carrying out privacy desensitization treatment on the initial user data to generate target user data.
Further, before the training of the constructed initial user portrait local model based on the acquired target user data, the client is specifically configured to:
judging whether a built initial user portrait local model needs to be trained or not according to a preset iteration number and a preset model convergence condition;
and if the constructed initial user portrait local model does not meet the preset iteration times and/or the preset model convergence condition, training the constructed initial user portrait local model.
Further, the client determines the constructed initial user portrait local model by:
obtaining sample user data and a sample user label in the sample user data; the sample user tags are used for characterizing real sample user portrait types of each sample user;
inputting the sample user data into an initial neural network for training, and determining a preset sample user portrait type corresponding to each sample user;
and when the loss value between the real sample user portrait type corresponding to each sample user and the preset sample user portrait type corresponding to the sample user is smaller than a preset threshold value, training is cut off, and a constructed initial user portrait local model is determined.
Further, the central server is specifically configured to:
performing parameter aggregation processing on the plurality of intermediate local network parameters, and determining aggregated network parameters;
and training the constructed initial user portrait global model according to the aggregated network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model.
Further, before the training of the constructed initial user representation global model according to the plurality of intermediate local network parameters sent by the plurality of clients, the central server is further configured to:
receiving connection requests sent by various clients;
determining whether each client meets a preset connection authority or not according to each connection request;
and if so, establishing connection with the client.
In a second aspect, an embodiment provided by the present application further provides a user representation-based user service method, which is applied to any client in any user representation-based user service system described in the first aspect, where the user representation determination method based on privacy computation includes:
acquiring target user data;
training the constructed initial user portrait local model based on the acquired target user data, and determining intermediate local network parameters of the updated initial user portrait local model;
and re-training the updated initial user portrait local model according to the intermediate global network parameters sent by the central server until the trained target user portrait local model and the target local network parameters corresponding to the target user portrait local model are determined, so that the central server can generate a target user portrait global model based on the target local network parameters.
In a third aspect, the present application provides an embodiment of a user portrait-based user service method, applied to a central server in a user portrait-based user service system as claimed in any one of the first aspect, where the user portrait determination method based on privacy computation includes:
acquiring a plurality of intermediate local network parameters sent by a plurality of clients;
training the constructed initial user portrait global model according to the plurality of intermediate local network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model;
and determining a target user portrait global model based on the initial user portrait global model after each target local network parameter is updated and sent by each client, so that each client can use the target user portrait global model to divide user types and recommend services to each target user.
An embodiment of the present application further provides an electronic device, including: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is operating, the machine-readable instructions when executed by the processor performing the steps of the user representation-based user service method of the third or fourth aspect as described above.
Embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the user portrait based user service method according to the third aspect or the fourth aspect.
Compared with the prior art, the user service system, the method and the electronic equipment based on the user portrait provided by the embodiment of the application have the advantages that intermediate local network parameters generated in the process of model training of each client are locally input into the initial user portrait global model in the central server for continuous iterative training, the updated intermediate global network parameters of the initial user portrait global model are sent to each client, the central server collects the installation and determination of the target user portrait global model through continuous interaction between each client and the central server, the purpose of privacy protection of client data is achieved by only uploading the network parameters of the model in each client without directly uploading internal data, and the safety of the data is improved.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
FIG. 1 is a block diagram illustrating a user service system based on user profiles according to an embodiment of the present application;
FIG. 2 is a flowchart of a user service method based on a user profile according to an embodiment of the present application;
FIG. 3 illustrates a second flowchart of a user service method based on a user profile according to an embodiment of the present application;
fig. 4 shows a schematic structural diagram of an electronic device according to an embodiment of the present application.
In the figure:
10-user service system based on user profile; 100-a client; 200-a central server; 400-an electronic device; 410-a processor; 420-a memory; 430-bus.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, as presented in the figures, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. Every other embodiment that can be obtained by a person skilled in the art without making creative efforts based on the embodiments of the present application falls within the protection scope of the present application.
First, an application scenario to which the present application is applicable will be described. The method and the device can be applied to the technical field of data processing.
Research shows that the traditional copy-type data circulation mode seriously leaks business privacy information, personal privacy information and the like, and cannot meet the requirements of privacy protection and data security of corresponding data of a user.
In addition, the traditional "replicated" data circulation method performs calculation at a data provider (i.e., a central server), and although data can not be out of the domain, calculation rules and calculation models of a business party (i.e., a client) are exposed, and thus business privacy of the business party is exposed.
In the prior art, the flow among different links of the data value chain is blocked, the work division cooperation is weak, and an effective closed loop is difficult to form.
The data value chain is a theoretical innovation for describing data value creation, divides data value creation activities into basic value activities and value-added activities, and achieves value creation of data and value addition in a transmission process through the value activities. The data value chain theory is different from the traditional value chain theory, and emphasizes that the value creation of data and the value increment in the transmission process are realized through the acquisition, transmission, storage, analysis and application of data on each node of the value chain.
Based on this, the embodiment of the application provides a user service system and method based on a user portrait, and an electronic device, which achieve the purpose of privacy protection of client data by directly uploading network parameters of models in each client, thereby improving the security of data.
Referring to fig. 1, fig. 1 is a block diagram illustrating a user service system based on a user representation according to an embodiment of the present disclosure. As shown in FIG. 1, a user representation-based user service system 10 provided by an embodiment of the present application includes a plurality of clients 100 and a central server 200.
In the above, the user service system for user representation is composed of a plurality of clients 100 and a central server 200, and the relationship between the plurality of clients 100 and the central server 200 is a usage but not limited to a usage of federal learning.
The core idea of the distributed machine Learning technology is that distributed model training is performed among a plurality of data sources with local data, and a global model based on virtual fusion data is constructed only by exchanging model parameters or intermediate results on the premise of not exchanging local individuals or sample data, so that balance between data privacy protection and data sharing calculation is realized, namely a new application paradigm of 'data available invisible' and 'data motionless model dynamic'.
Each client 100 is configured to train the constructed initial user portrait local model based on the acquired target user data, determine an intermediate local network parameter of the updated initial user portrait local model, and send the intermediate local network parameter to the central server 200; and training the updated initial user portrait local model again according to the intermediate global network parameters sent by the central server 200 until the trained target user portrait local model is determined, and sending target local network parameters corresponding to the target user portrait local model to the central server 200.
Here, in the process of training the target user portrait global model, each client 100 trains the constructed initial user portrait local model according to the acquired target user data, and generates the updated initial user portrait local model and the updated intermediate local network parameters of the initial user portrait local model, so that the constructed initial user portrait global model in the central server 200 can train the user portrait global model according to the intermediate local network parameters.
After receiving the intermediate global network parameters sent by the central server 200, the client 100 continues to train the updated initial user portrait local model, continuously updates the iterative user portrait local model, and continuously outputs the intermediate local network parameters to the central server 200 until the trained target user portrait local model is determined, and sends the target local network parameters corresponding to the target user portrait local model to the central server 200, so as to complete the interaction between the client 100 and the central server 200.
In this way, the client 100 only sends the local network parameters to the central server 200, and does not send the sample data set of its local training, i.e. does not share the original local data, so that the data provider (client 100) "zero-information ex-warehouse", the data demander (central server 200) "zero-privacy leakage", and further the privacy protection for the target user is realized.
In the above, since the number of the clients 100 is multiple, the collected user data may have more diversified dimensions, and may include, but not be limited to, user data with dimensions including government affairs, operators, the internet, and the like, and further, the embodiment provided by the present application may perform type analysis on user portrayal for users other than the clients 100, and the embodiment provided by the present application may replace a trusted third party in federal learning by a block chain technique, and is responsible for saving and updating a user portrayal global model, coordinating and configuring each client 100 to train a built initial user portrayal local model.
Optionally, the client 100 determines the target user data by the following sub-steps:
and substep 1, acquiring initial user data.
In this step, the initial user data in the client 100 includes, but is not limited to, the type of data that the user fills in the client 100, such as the user's identity data, the user's purchase data, and the user's internet usage data, which are personal privacy data of the user.
And substep 2, carrying out privacy desensitization processing on the initial user data to generate target user data.
In this step, the embodiment provided by the application may add noise to the initial user data by using a self-adaptive differential privacy method, dynamically adjust the clipping threshold according to the priori knowledge, desensitize the privacy of the initial user data, and generate the target user data.
Here, the desensitization process is actually to obfuscate and hide the data.
The embodiment provided by the present application explains a way of adding noise by a specific embodiment one:
in the t-th round training initial user portrait local model, user U i Clipping gradient g at local client 100 i,t Adding noise and dynamically adjusting clipping threshold C t Can be represented as follows:
thus, after the privacy desensitization processing is performed on the initial user data, target user data is generated.
Optionally, before the training of the constructed initial user portrait local model based on the acquired target user data, the client 100 is specifically configured to:
1) And judging whether the constructed initial user portrait local model needs to be trained or not according to the preset iteration times and the preset model convergence condition.
Here, if the initial user portrait local model has reached the preset iteration number and the initial user portrait local model has reached the preset model convergence condition during the initial training, the constructed initial user portrait local model does not need to be trained.
2) And if the constructed initial user portrait local model does not meet the preset iteration times and/or the preset model convergence condition, training the constructed initial user portrait local model.
Here, if the constructed initial user portrait local model does not satisfy the preset iteration number and/or the preset model convergence condition, the client 100 actively initiates a connection request with the central server 200, receives the intermediate global network parameter sent by the central server 200 after the central server 200 agrees, and starts to train in the constructed initial user portrait local model.
Optionally, the client 100 determines the constructed initial user portrait partial model by the following sub-steps:
substep 1, obtaining sample user data and a sample user label in the sample user data; the sample user tags are used to characterize a true sample user representation type for each of the sample users.
In this step, the sample user data tag may be represented as P = ∑ tone<U 1 ,C 1 >,<U 2 ,C 2 >,…,<U m ,C n >}. Wherein, U i For unstructured data describing user information, C i And (3) representing the type label for the user for the real sample user of the user, wherein m is the total number of documents, and n is the total number of categories.
Here, unstructured data is data that has an irregular or incomplete data structure, does not have a predefined data model, and is not conveniently represented by a database two-dimensional logical table. Including office documents, text, pictures, HTML, various forms, images, and audio/video information in all formats.
And substep 2, inputting the sample user data into an initial neural network for training, and determining a preset sample user portrait type corresponding to each sample user.
In this step, the default sample user profile type may be denoted as D i ＝{U i ,<f 1 ,p 1 >,<f 2 ,p 2 >,…,<f k ,p k >In which f i Preset sample user portrait type, p, representing the ith user i Representing a user U i True sample user representation type f with user i The probability of (c).
And 3, when the loss value between the real sample user portrait type corresponding to each sample user and the preset sample user portrait type corresponding to the sample user is smaller than a preset threshold value, training is cut off, and a constructed initial user portrait local model is determined.
In this step, the softmax regression algorithm may be used in the embodiments provided herein, but is not limited to use, to train the initial user portrait local model.
Here, the softmax regression algorithm itself can be used as a learning algorithm to optimize the classification result, which is only an additional processing layer in the neural network, and is used to change the output of the neural network into a probability distribution output.
The central server 200 is configured to train the constructed initial user portrait global model according to the plurality of intermediate local network parameters sent by the plurality of clients 100, determine intermediate global network parameters of the updated initial user portrait global model, and send the intermediate global network parameters to each of the clients 100; and determining a target user portrait global model based on each target local network parameter sent by each client 100 and the updated initial user portrait global model, so that each client 100 uses the target user portrait global model to divide user types of each target user and recommend services.
Here, the central server 200 trains an initial user portrait global model and an updated initial user portrait global model using a Federated learning deep neural network method (fed iterative) based on average iteration, determines a target user portrait local model, and after the training is completed, and distributes the target user portrait local model to each client 100, the client 100 uses the local data and the distribution of interest tags of each category of target users output by the target user portrait local model as the user portrait of the group.
In the above, the intermediate global network parameters obtained by the training of the central server 200 may be distributed to the clients 100 that need to be trained in the next iteration.
The central server 200 includes the target user data of each client 100, and the obtaining manner of the target user data of each client 100 in the central server 200 includes but is not limited to: after the central server 200 trains the initial neural network according to the training user data of itself, the target user data in the client 100 is mapped into the feature space vector of the training user data of itself, and the target user data in the central server 200 only stores the identity attribute of the user.
Here, the process of the user service system 10 for user type classification and service recommendation for each target user based on user profile is described as a specific embodiment:
dividing sample user labels of marketing touch points into four categories of understanding, interested, purchasing clients and faithful clients, cleaning, extracting features and labeling sample user data on an Industrial Internet platform (INDICES), using the processed sample users and the corresponding sample user labels as participants of a client 100 for federal learning, randomly extracting a certain amount of sample user labels for federal learning, training the user types of the sample users by adopting a softmax regression portrait algorithm on the client 100, uploading intermediate local network parameters of each training round to a central server 200 for parameter aggregation and updating, and obtaining a global user portrait global model so that each client 100 can use the target user global model to divide the user types of each target user and recommend services, thereby improving the operation efficiency.
Optionally, the central server 200 is specifically configured to:
1) And performing parameter aggregation processing on the plurality of intermediate local network parameters, and determining the aggregated network parameters.
Here, the parameter aggregation manner in the embodiments provided in the present application includes, but is not limited to, a manner including averaging after summing.
2) And training the constructed initial user portrait global model according to the aggregated network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model.
Here, the inputs that determine the updated initial user portrait global model are the intermediate local network parameters w involved in federal learning 1 ,w 2 ,…,w k (ii) a And determining the output of the updated initial user portrait global model as an intermediate global network parameter.
The specific way to determine the updated initial user portrait global model is as follows:
initialize w0; v/initialize local network parameters;
for T =1,2,. Multidot., tdo; v/each communication in t iterations;
m ← max (γ × N, 1), γ ∈ [0,1]; the number of target users is/m, and N is the total number of categories of the target users;
ct ← random set of m set; v, randomly selecting target users in the current round and putting the target users into a federal learning set;
for C belongs to Ct do; // each target user for federal learning;
end for；；
end for；
optionally, before the training of the constructed initial user representation global model according to the plurality of intermediate local network parameters sent by the plurality of clients 100, the central server 200 is further configured to:
1) A connection request sent by each client 100 is received.
2) And determining whether each client 100 meets a preset connection authority or not according to each connection request.
Here, the preset connection authority includes, but is not limited to, a threshold of the number of times of connection of the client 100 to the central server 200, and if the client 100 is the next client 100 exceeding the threshold of the number of times of connection, the client 100 cannot establish connection with the central server 200 in the current round, and can establish connection with the central server 200 in the next round.
3) If yes, a connection with the client 100 is established.
Compared with the prior art, the user service system 10 based on the user portrait provided by the embodiment of the application inputs the initial user portrait global model in the central server 200 for continuous iterative training by using the intermediate local network parameters generated in the local model training process of each client 100, sends the updated intermediate global network parameters of the initial user portrait global model to each client 100, completes the installation and determination of the target user portrait global model summarized by the central server 200 through the continuous interaction between each client 100 and the central server 200, and realizes the purpose of privacy protection of the data of the client 100 by only uploading the network parameters of the model in each client 100 without directly uploading the internal data, thereby improving the security of the data.
The embodiment provided by the application realizes privacy calculation by carrying out privacy desensitization processing on user data, ensures the security of the user data, and can realize accurate customized service recommendation and marketing service schemes for different target users by using the target user portrait global model.
The privacy calculation in the embodiment provided by the application can enhance the protection of target user data and reduce the risk of data leakage. Compared with the traditional data security means, the privacy calculation in the embodiment provided by the application maximizes the user data value as much as possible on the premise of ensuring the user data security.
Referring to fig. 2, fig. 2 is a flowchart illustrating a user service method based on a user profile according to an embodiment of the present application. As shown in fig. 2, a user service method based on a user profile provided in an embodiment of the present application includes the following steps:
s201, acquiring target user data.
Optionally, the target user data is determined by:
initial user data is obtained.
And carrying out privacy desensitization processing on the initial user data to generate target user data.
S202, training the constructed initial user portrait local model based on the acquired target user data, and determining intermediate local network parameters of the updated initial user portrait local model.
Optionally, before the training of the constructed initial user representation local model based on the acquired target user data, the user service method based on the user representation further includes:
and judging whether the constructed initial user portrait local model needs to be trained or not according to the preset iteration times and the preset model convergence condition.
And if the constructed initial user portrait local model does not meet the preset iteration times and/or the preset model convergence condition, training the constructed initial user portrait local model.
Optionally, the constructed initial user portrait partial model is determined by:
obtaining sample user data and a sample user label in the sample user data; the exemplar user tags are used to characterize real exemplar user representation types for each of the exemplar users.
Inputting the sample user data into an initial neural network for training, and determining the type of the preset sample user portrait corresponding to each sample user.
And when the loss value between the real sample user portrait type corresponding to each sample user and the preset sample user portrait type corresponding to the sample user is smaller than a preset threshold value, training is cut off, and a constructed initial user portrait local model is determined.
S203, retraining the updated initial user portrait local model according to the intermediate global network parameters sent by the central server until the trained target user portrait local model and the target local network parameters corresponding to the target user portrait local model are determined, so that the central server can generate a target user portrait global model based on the target local network parameters.
Compared with the prior art, the user service method based on the user portrait provided by the embodiment of the application has the advantages that the intermediate local network parameters generated in the process of model training of each client side are locally input into the initial user portrait global model in the central server for continuous iterative training, the updated intermediate global network parameters of the initial user portrait global model are sent to each client side, installation and determination of the central server for summarizing the target user portrait global model are completed through continuous interaction between each client side and the central server, the purpose of privacy protection of client side data is achieved through a mode that only the network parameters of the model in each client side are uploaded, and internal data are not directly uploaded, and accordingly data security is improved.
The embodiment provided by the application realizes privacy calculation by carrying out privacy desensitization processing on user data, ensures the security of the user data, and can realize accurate customized service recommendation and marketing service schemes for different target users by using the target user portrait global model.
The privacy calculation in the embodiment provided by the application can enhance the protection of target user data and reduce the risk of data leakage. Compared with the traditional data security means, the privacy calculation in the embodiment provided by the application maximizes the user data value as much as possible on the premise of ensuring the user data security.
Referring to fig. 3, fig. 3 is a second flowchart of a user service method based on a user representation according to an embodiment of the present application. As shown in fig. 3, the user profile-based user service method includes the steps of:
s301, acquiring a plurality of intermediate local network parameters sent by a plurality of clients.
S302, training the constructed initial user portrait global model according to the plurality of intermediate local network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model.
Optionally, the training the constructed initial user portrait global model according to the plurality of intermediate local network parameters, and determining the updated intermediate global network parameters of the initial user portrait global model includes:
and performing parameter aggregation processing on the plurality of intermediate local network parameters, and determining the aggregated network parameters.
And training the constructed initial user portrait global model according to the aggregated network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model.
Optionally, before the training of the constructed initial user portrait global model according to the plurality of intermediate local network parameters sent by the plurality of clients, the user service method based on user portraits further includes:
and receiving connection requests sent by the clients.
And determining whether each client meets a preset connection authority or not according to each connection request.
And if so, establishing connection with the client.
And S303, determining a target user portrait global model based on the initial user portrait global model which is sent by each client and is updated by each target local network parameter, so that each client can use the target user portrait global model to divide user types of each target user and recommend services.
Compared with the prior art, the user service method based on the user portrait provided by the embodiment of the application has the advantages that the intermediate local network parameters generated in the process of model training of each client side are locally input into the initial user portrait global model in the central server for continuous iterative training, the updated intermediate global network parameters of the initial user portrait global model are sent to each client side, installation and determination of the central server for summarizing the target user portrait global model are completed through continuous interaction between each client side and the central server, the purpose of privacy protection of client side data is achieved through a mode that only the network parameters of the model in each client side are uploaded, and internal data are not directly uploaded, and accordingly data security is improved.
The embodiment provided by the application realizes privacy calculation by carrying out privacy desensitization processing on user data, ensures the security of the user data, and can realize accurate customized service recommendation and marketing service schemes for different target users by using the target user portrait global model.
The privacy calculation in the embodiment provided by the application can enhance the protection of target user data and reduce the risk of data leakage. Compared with the traditional data security means, the privacy calculation in the embodiment provided by the application maximizes the user data value as much as possible on the premise of ensuring the user data security.
Referring to fig. 4, fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure. As shown in fig. 4, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.
The memory 420 stores machine-readable instructions executable by the processor 410, when the electronic device 400 runs, the processor 410 communicates with the memory 420 through the bus 430, and when the machine-readable instructions are executed by the processor 410, the steps of the user service method based on the user representation in the method embodiments shown in fig. 2 and fig. 3 may be executed.
An embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the user service method based on a user representation in the method embodiments shown in fig. 2 and fig. 3 may be executed.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the units into only one type of logical function may be implemented in other ways, and for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solutions of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used to illustrate the technical solutions of the present application, but not to limit the technical solutions, and the scope of the present application is not limited to the above-mentioned embodiments, although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present application and are intended to be covered by the appended claims. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A user profile-based user service system, comprising a plurality of clients and a central server;
each client is used for training the constructed initial user portrait local model based on the acquired target user data, determining intermediate local network parameters of the updated initial user portrait local model, and sending the intermediate local network parameters to the central server; training the updated initial user portrait local model again according to the intermediate global network parameters sent by the central server until a trained target user portrait local model is determined, and sending target local network parameters corresponding to the target user portrait local model to the central server;
the central server is used for training the constructed initial user portrait global model according to the plurality of intermediate local network parameters sent by the plurality of clients, determining the intermediate global network parameters of the updated initial user portrait global model, and sending the intermediate global network parameters to the clients; and determining a target user portrait global model based on each target local network parameter sent by each client and the updated initial user portrait global model so that each client can use the target user portrait global model to divide user types and recommend services for each target user.
2. The user representation-based user service system of claim 1, wherein the client determines the target user data by:
acquiring initial user data;
and carrying out privacy desensitization treatment on the initial user data to generate target user data.
3. The user representation-based user service system of claim 1, wherein prior to the training of the constructed initial user representation local model based on the obtained target user data, the client is specifically configured to:
judging whether a built initial user portrait local model needs to be trained or not according to a preset iteration number and a preset model convergence condition;
and if the constructed initial user portrait local model does not meet the preset iteration times and/or the preset model convergence condition, training the constructed initial user portrait local model.
4. The user representation-based user service system of claim 1, wherein the client determines the constructed initial user representation local model by:
acquiring sample user data and a sample user label in the sample user data; the sample user tags are used for characterizing real sample user portrait types of each sample user;
inputting the sample user data into an initial neural network for training, and determining a preset sample user portrait type corresponding to each sample user;
and when the loss value between the real sample user portrait type corresponding to each sample user and the preset sample user portrait type corresponding to the sample user is smaller than a preset threshold value, training is cut off, and a constructed initial user portrait local model is determined.
5. The user representation-based user service system of claim 1, wherein the central server is specifically configured to:
performing parameter aggregation processing on the plurality of intermediate local network parameters, and determining aggregated network parameters;
and training the constructed initial user portrait global model according to the aggregated network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model.
6. The user representation-based user service system of claim 1, wherein prior to the training of the constructed initial user representation global model based on the plurality of intermediate local network parameters sent by the plurality of clients, the central server is further configured to:
receiving connection requests sent by all clients;
determining whether each client meets a preset connection authority or not according to each connection request;
and if so, establishing connection with the client.
7. A user representation-based user service method, for use at any client in a user representation-based user service system as claimed in any one of claims 1 to 6, the privacy-computation-based user representation determination method comprising:
acquiring target user data;
training the constructed initial user portrait local model based on the acquired target user data, and determining intermediate local network parameters of the updated initial user portrait local model;
and re-training the updated initial user portrait local model according to the intermediate global network parameters sent by the central server until the trained target user portrait local model and the target local network parameters corresponding to the target user portrait local model are determined, so that the central server can generate a target user portrait global model based on the target local network parameters.
8. A user profile based user service method for use in a central server in a user profile based user service system as claimed in any one of claims 1 to 6, said privacy calculation based user profile determination method comprising:
acquiring a plurality of intermediate local network parameters sent by a plurality of clients;
training the constructed initial user portrait global model according to the plurality of intermediate local network parameters, and determining the intermediate global network parameters of the updated initial user portrait global model;
and determining a target user portrait global model based on the initial user portrait global model after each target local network parameter is updated and sent by each client, so that each client can use the target user portrait global model to divide user types and recommend services to each target user.
9. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor for communicating with the memory over the bus when an electronic device is operating, the machine-readable instructions when executed by the processor performing the steps of the user representation-based user service method of claim 7 or the user representation-based user service method of claim 8.
10. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, performs the steps of the user representation-based user service method of claim 7 or the user representation-based user service method of claim 8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211526982.9A CN115718930A (en) | 2022-11-30 | 2022-11-30 | User service system and method based on user portrait and electronic equipment |
| PCT/CN2023/134756 WO2024114640A1 (en) | 2022-11-30 | 2023-11-28 | User portrait-based user service system and method, and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211526982.9A CN115718930A (en) | 2022-11-30 | 2022-11-30 | User service system and method based on user portrait and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115718930A true CN115718930A (en) | 2023-02-28 |
Family
ID=85257189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211526982.9A Pending CN115718930A (en) | 2022-11-30 | 2022-11-30 | User service system and method based on user portrait and electronic equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115718930A (en) |
| WO (1) | WO2024114640A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117829968A (en) * | 2024-03-06 | 2024-04-05 | 南京数策信息科技有限公司 | Service product recommendation method, device and system based on user data analysis |
| WO2024114640A1 (en) * | 2022-11-30 | 2024-06-06 | 北京航天数据股份有限公司 | User portrait-based user service system and method, and electronic device |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020062262A1 (en) * | 2018-09-30 | 2020-04-02 | Shanghai United Imaging Healthcare Co., Ltd. | Systems and methods for generating a neural network model for image processing |
| CN112116008B (en) * | 2020-09-18 | 2024-07-05 | 平安科技（深圳）有限公司 | Processing method of target detection model based on intelligent decision and related equipment thereof |
| CN112329940A (en) * | 2020-11-02 | 2021-02-05 | 北京邮电大学 | Personalized model training method and system combining federal learning and user portrait |
| CN112416986B (en) * | 2020-11-23 | 2024-03-29 | 中国科学技术大学 | User portrait realizing method and system based on hierarchical personalized federal learning |
| CN113283185B (en) * | 2021-07-23 | 2021-11-12 | 平安科技（深圳）有限公司 | Federal model training and client imaging method, device, equipment and medium |
| CN115718930A (en) * | 2022-11-30 | 2023-02-28 | 北京航天数据股份有限公司 | User service system and method based on user portrait and electronic equipment |
-
2022
- 2022-11-30 CN CN202211526982.9A patent/CN115718930A/en active Pending
-
2023
- 2023-11-28 WO PCT/CN2023/134756 patent/WO2024114640A1/en unknown
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024114640A1 (en) * | 2022-11-30 | 2024-06-06 | 北京航天数据股份有限公司 | User portrait-based user service system and method, and electronic device |
| CN117829968A (en) * | 2024-03-06 | 2024-04-05 | 南京数策信息科技有限公司 | Service product recommendation method, device and system based on user data analysis |
| CN117829968B (en) * | 2024-03-06 | 2024-05-31 | 南京数策信息科技有限公司 | Service product recommendation method, device and system based on user data analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2024114640A1 (en) | 2024-06-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11023107B2 (en) | Virtual teller systems and methods | |
| CN110162717B (en) | Method and device for recommending friends | |
| Tang et al. | Collaborative web service quality prediction via exploiting matrix factorization and network map | |
| WO2024114640A1 (en) | User portrait-based user service system and method, and electronic device | |
| WO2018145590A1 (en) | Emoticon picture recommendation method and device, server cluster and storage medium | |
| Folorunso et al. | A fuzzy expert system to Trust-Based Access Control in crowdsourcing environments | |
| CN112534431B (en) | Improving security of cryptographically protected resources based on publicly available data | |
| CN113254804B (en) | Social relationship recommendation method and system based on user attributes and behavior characteristics | |
| CN113505882A (en) | Data processing method based on federal neural network model, related equipment and medium | |
| CN111709051A (en) | Data processing method, device and system, computer storage medium and electronic equipment | |
| Zhou et al. | Measuring web service security in the era of Internet of Things | |
| CN110889759A (en) | Credit data determination method, device and storage medium | |
| Shafik et al. | Recommendation system comparative analysis: internet of things aided networks | |
| CN113780321A (en) | Picture auditing method and system, electronic equipment and computer readable medium | |
| CN111353554B (en) | Method and device for predicting missing user service attributes | |
| CN113761350A (en) | Data recommendation method, related device and data recommendation system | |
| CN112182399A (en) | Multi-party security calculation method and device for federated learning | |
| Saputra et al. | Federated learning framework with straggling mitigation and privacy-awareness for AI-based mobile application services | |
| Ali-Eldin et al. | A risk evaluation approach for authorization decisions in social pervasive applications | |
| Chamoso et al. | Social computing for image matching | |
| Vahabli et al. | A novel trust-based access control for social networks using fuzzy systems | |
| CN117094773A (en) | Online migration learning method and system based on blockchain privacy calculation | |
| CN113112312B (en) | Method, apparatus and computer-readable storage medium for generating a model for a user | |
| CN116415064A (en) | Training method and device for double-target-domain recommendation model | |
| Zupancic et al. | TACO: a novel method for trust rating subjectivity elimination based on Trust Attitudes COmparison |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |