CN116228326A

CN116228326A - Data processing method, system, equipment and medium

Info

Publication number: CN116228326A
Application number: CN202310099140.8A
Authority: CN
Inventors: 朱洪渊; 牛也; 邓云辉
Original assignee: Alibaba Cloud Computing Ltd
Current assignee: Alibaba Cloud Computing Ltd
Priority date: 2023-01-20
Filing date: 2023-01-20
Publication date: 2023-06-06

Abstract

The embodiment of the application provides a data processing method, a system, equipment and a medium. The method comprises the following steps: and acquiring equipment space-time data related to the terminal equipment, data to be promoted and user data of a plurality of different users interacting with the terminal equipment. And carrying out feature extraction on the equipment space-time data, the data to be promoted and the user data by using a feature extractor to obtain equipment flow features, promotion data features and user features represented by the designated subgraph. And utilizing a recommendation model to determine target popularization data aiming at the terminal equipment from a plurality of data to be promoted based on the fused multi-mode characteristics. The method is used for collecting user data experienced by the terminal equipment with larger user mobility, and accurately extracting terminal flow characteristics of the terminal equipment by utilizing the corresponding characteristic extractor, so that the user characteristics of the current user can be integrated, and the promotion data characteristics of the data to be promoted and the equipment flow characteristics comprehensively evaluate target promotion data recommended to the user through the terminal equipment.

Description

Data processing method, system, equipment and medium

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a data processing method, system, device, and medium.

Background

With the development of computer technology, a lot of convenience can be provided for the daily life of users. Such as a vending machine.

In the prior art, vending machines are typically installed at a business location for use by different users. For example, a user may go to a mall to consume the beverage, first watch a movie, then go to a vending machine after watching the movie, and want to buy the beverage, at this time, an advertisement of a certain garment or an advertisement of a beverage of a certain brand may be played on a display screen of the vending machine. Because the vending machine provides beverages for a plurality of different users every day, the actual demand of each user is not known, and therefore, the advertisement content displayed through the display screen does not necessarily meet the demand of the user, and accurate popularization cannot be realized.

Disclosure of Invention

In order to solve or improve the problems existing in the prior art, embodiments of the present application provide a data processing method, system, device, and medium.

In a first aspect, in one embodiment of the present application, a data processing method is provided. The method is applied to the server side, and comprises the following steps:

acquiring equipment space-time data related to terminal equipment, data to be promoted and user data of a plurality of different users interacting with the terminal equipment, wherein the user data are desensitization data or statistical data, and the terminal equipment is equipment for providing services for unfixed users;

Extracting the characteristics of the equipment space-time data, the data to be promoted and the user data by using a characteristic extractor to obtain equipment flow characteristics, promotion data characteristics and user characteristics represented by a designated sub-graph;

and carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model so as to utilize the recommendation model to determine target promotion data aiming at the current user in interaction with the terminal equipment from a plurality of data to be promoted.

In a second aspect, in one embodiment of the present application, there is provided a data processing method applied to a terminal device, where the terminal device is a device that provides a service to a non-stationary user, the method including:

responding to an interaction request of a first user through a client, and acquiring a first user identification;

the first user identification and the equipment identification of the terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to extract features of equipment space-time data, data to be promoted and user data, and equipment flow features, promotion data features and user features represented by a designated subgraph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data;

And receiving target promotion data for the terminal equipment, which are determined by the server side from a plurality of data to be promoted by utilizing the recommendation model.

In a third aspect, in one embodiment of the present application, there is provided a data processing method applied to a user portable device, the method including:

after sending an interaction request to terminal equipment, receiving feedback information carrying equipment identification;

the method comprises the steps that a first user identifier and a device identifier of terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to conduct feature extraction on device space-time data, data to be promoted and user data, and device flow features, promotion data features and user features represented by a designated subgraph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data;

In a fourth aspect, in an embodiment of the present application, a data processing system is provided, which includes a data processing method of a server side according to the first aspect and a data processing method of a terminal device according to the second aspect.

In a fifth aspect, in an embodiment of the present application, a data processing system is provided, which includes a data processing method of the server side according to the first aspect and a data processing method of the user portable device according to the third aspect.

In a sixth aspect, in one embodiment of the present application, there is provided a cloud server, including a memory and a processor; wherein,,

the memory is used for storing programs;

the processor is coupled to the memory, and is configured to execute the program stored in the memory, so as to implement the data processing method according to the first aspect.

In a seventh aspect, in one embodiment of the present application, there is provided a non-transitory machine-readable storage medium having stored thereon executable code, which when executed by a processor of an electronic device, causes the processor to perform the data processing method according to the first aspect, or the data processing method according to the second aspect, or the data processing method according to the third aspect.

According to the technical scheme provided by the embodiment of the application, in an online scene, some terminal equipment does not provide service for a unique user, but provides service for equipment of an unfixed user. The server side can directly or indirectly provide promotion data for the terminal equipment, so that the promotion data better meets the requirement of a first user currently interacting with the terminal equipment, the server side obtains more comprehensive data, the data comprise equipment space-time data, data to be promoted and user data of a plurality of different users interacting with the terminal equipment (including the user data currently interacting and interacted), and the user data is desensitized data or statistical data. Further, the feature extraction is performed by a feature extractor, and the corresponding features are obtained as follows: device flow characteristics, promotional data characteristics, and user characteristics represented by the specified sub-graph. And further fusing the characteristics corresponding to different data types, and determining target popularization data suitable for the user interacting with the terminal equipment at present through a recommendation model. The users with the interaction behavior with the terminal equipment often have the same or similar characteristics, so that the appointed sub-graph representation can be selected to serve as the equipment flow characteristic under the condition of considering the calculation efficiency and the characteristic extraction effect, the common characteristics of the users with the interaction behavior with the terminal equipment can be comprehensively represented, and the calculation amount can be effectively reduced. In addition, user data experienced by the terminal equipment with larger user mobility is collected, the terminal flow characteristics of the terminal equipment are accurately extracted by utilizing the corresponding characteristic extractor, the user characteristics of the current user can be further integrated, the promotion data characteristics of the data to be promoted and the equipment flow characteristics comprehensively evaluate the target promotion data recommended to the user through the terminal equipment, the accuracy of the terminal equipment in data promotion can be effectively improved, and the user can obtain the promotion data required by the user on strange terminal equipment.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief description will be given below of the drawings that are needed in the embodiments or the prior art descriptions, and it is obvious that the drawings in the following description are some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.

FIGS. 1a and 1b are schematic illustrations of data processing systems according to embodiments of the present application;

FIG. 2 is a flow chart illustrating a data processing method according to an embodiment of the present application;

FIG. 3 is a schematic diagram of feature fusion as illustrated in an embodiment of the present application;

fig. 4 is a schematic flow chart of weight-based computation fusion according to an embodiment of the present application;

FIG. 5 is a flowchart illustrating another data processing method according to an embodiment of the present disclosure;

FIG. 6 is a flowchart illustrating another data processing method according to an embodiment of the present disclosure;

FIG. 7 is a schematic diagram of a data processing apparatus according to an embodiment of the present disclosure;

FIG. 8 is a schematic diagram of another data processing apparatus according to an embodiment of the present disclosure;

FIG. 9 is a schematic diagram of a data processing apparatus according to another embodiment of the present disclosure;

fig. 10 is a schematic structural diagram of an electronic device according to an embodiment of the present application;

fig. 11 is a schematic structural diagram of another electronic device according to an embodiment of the present application;

fig. 12 is a schematic structural diagram of still another electronic device according to an embodiment of the present application.

Detailed Description

In order to enable those skilled in the art to better understand the present application, the following description will make clear and complete descriptions of the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application.

In some of the flows described in the specification, claims, and drawings described above, a plurality of operations occurring in a particular order are included, and the operations may be performed out of order or concurrently with respect to the order in which they occur. The sequence numbers of operations such as 101, 102, etc. are merely used to distinguish between the various operations, and the sequence numbers themselves do not represent any order of execution. In addition, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first" and "second" herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, and are not limited to the "first" and the "second" being different types. Furthermore, the embodiments described below are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

In the conventional off-line retail scenario, the fluidity of the customer is relatively high, and the off-line merchant cannot accurately know the consumption habit, hobbies and the like of each user. When the online popularization is carried out, only the general popularization without pertinence can be adopted, for example, the advertisement of a certain beverage and the advertisement of a certain food brand are repeatedly played through a large screen on the automatic vending machine without pertinence. The popularization mode has weak pertinence, low click rate and poor popularization effect. Therefore, a scheme capable of realizing accurate popularization of terminal equipment for different users is needed.

Term interpretation:

multi-modal learning (Multimodal Learning, MML) performs feature fusion learning based on a plurality of different modal features.

The artificial intelligence internet of things (Artificial intelligence & Internet of Things, AIOT) is integrated with the AI technology and the IoT technology, mass data from different dimensions are generated and collected through the internet of things and stored at the cloud end and the edge end, and then the everything is dataized and the everything is intelligent through big data analysis and artificial intelligence in a higher form.

The technical scheme realized by the application will be explained below with reference to specific embodiments.

Fig. 1a and 1b are schematic illustrations of a data processing system according to an embodiment of the present application. As can be seen from fig. 1a and fig. 1b, the server side (in practical application, there may be a plurality of server sides capable of communicating with each other, each server side has a specific function, and different server sides can mutually transmit data), the user portable device and the terminal device.

The terminal device is a device without a fixed service object, in other words the terminal device may serve different users at a fixed location or at different locations. Because the terminal equipment often provides services for different users, but does not belong to a certain user, the mobility of personnel facing the terminal equipment is very high, the terminal equipment cannot acquire detailed information and interest of the certain user, and further in the traditional scheme, the terminal equipment cannot realize accurate data popularization aiming at different users. The terminal device may have a fixed location (e.g., vending machine, autonomous checkout device, manual checkout device, etc.), or may be movable as desired by the customer (e.g., a sharing vehicle).

The server side in the scheme is connected with the client side (the client side can be arranged on the portable equipment of the user, such as a mobile phone) of the user through a network, can provide services for the user at any time and any place, knows basic information of the user (the user needs to submit some basic information when registering), and the user also needs to provide support when using certain functions of the client side, and can also know behavior information of the user. Such as identity information of the user, historical consumption information, etc. And further, the basic information of the user can be utilized to know the consumption habit, hobbies and the like of the user.

The user portable device in the scheme can be, for example, a mobile phone, a watch, a computer and the like, can be provided with the electronic device of the client, and can perform data transmission with the server. Generally, user portable devices are personal. A user portable device may have installed thereon a client supporting various application functions. The server side described above provides service support for various application functions of the client side. Therefore, various actions realized by the user through the client are saved to the server, and the user characteristics obtained based on the user data are more accurate.

Fig. 1a and 1b are two functionally similar systems. The terminal device in fig. 1a may be an AIOT device, and may be directly connected to the server through a network and perform data transmission, when a first user interacts with the terminal device, the obtained user identifier of the first user and the device identifier of the terminal device may be sent to the server through the terminal device, after the server determines that the target promotion data is applicable to the first user, the server may directly send the target promotion data to the terminal device, and then the terminal device displays or plays the target promotion data to the first user.

In fig. 1b, the terminal device does not have the capability of directly networking with the server side, or is temporarily unable to networking with the server side, and may perform data transmission with the server side through the client side. And the obtained equipment identifier, the first user identifier and the like are sent to the server side by the user portable equipment. The server side obtains target promotion data after analysis and sends the target promotion data to the client side. And then the portable device of the user shows or plays the first user. If the client and the terminal equipment are kept connected in a Bluetooth mode or the like, the user portable equipment provided with the client can transmit the received target promotion data to the terminal equipment through a Bluetooth channel, and then the terminal equipment displays or plays the target promotion data to the first user. Or the portable device and the terminal device of the user can be displayed or played simultaneously.

Fig. 2 is a schematic flow chart of a data processing method according to an embodiment of the present application. The execution subject of the method may be a server side (e.g., cloud server). From fig. 2, it can be seen that the method specifically comprises the following steps:

201: acquiring equipment space-time data related to terminal equipment, data to be promoted and user data of a plurality of different users interacting with the terminal equipment, wherein the user data are desensitized data or statistical data, and the terminal equipment is equipment for providing services for unfixed users.

202: and carrying out feature extraction on the equipment space-time data, the data to be promoted and the user data by using a feature extractor to obtain equipment flow features, promotion data features and user features represented by the designated subgraph.

203: and carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model so as to utilize the recommendation model to determine target promotion data aiming at the current user in interaction with the terminal equipment from a plurality of data to be promoted.

The data to be promoted here may be advertisement data such as advertisement type, commodity color, shape, etc. The data to be promoted can be stored in a centralized manner at the server end, or can be stored at the server end and the equipment terminal at the same time. Although there are a lot of data to be promoted, the consumption habits of each user may be different. For example, different users have different eating habits. Therefore, target popularization data required by the first user currently interacting with the terminal equipment need to be accurately found from the data to be promoted, so that accurate popularization is realized.

It should be noted that, when determining the data to be promoted, the data to be promoted is to be related to the device attribute of the terminal device (for example, whether the terminal device is a mobile device (for example, a self-service vending machine or a shared vehicle), the service type provided by the device (for example, vending goods or providing shoe-cleaning services)), the device location (for example, in a mall, a school, a hospital or a park), and the data to be promoted (merchant) in the vicinity of the device location, or the data to be promoted related to the same location as the device location. For example, if a user purchases a beverage on a self-service vending machine located in the market a, the determined plurality of data to be promoted are data to be promoted corresponding to a plurality of merchants (such as catering and clothing) also located in the market a. For another example, if a user prepares to ride a shared bicycle or rent a shared bicycle in an downstairs square located in the market a to go to the market B, the determined plurality of data to be promoted are data to be promoted corresponding to a plurality of merchants (such as catering and clothing) in the market B to which the user will go.

The terminal device interacts with the user, which can be understood that the user interacts with the terminal device through the user portable device, and the interaction mode can be that the user portable device scans the two-dimensional code provided by the terminal device or the terminal device scans the two-dimensional code provided by the user portable device. During the interaction, data (e.g., device identification, user identification) transmission or payment operations are performed. The user data referred to herein is desensitized data or statistical data subjected to desensitization processing, for example, the following user's original data is subjected to desensitization processing: including user basic identity information, and consumer behavior performed by the user via the client (e.g., restaurant consumer behavior by the user via the client, online shopping behavior by the client), purchase merchandise types, restaurant flavor types, and the like. In addition, the method also comprises data related to the interaction of the user and the terminal equipment, including interaction time, interaction results and feedback results on promotion data (such as whether the user clicks on the promotion data). The device space-time data can be position data of the device and time data of interaction between different users and the device. In order to ensure data security, the original data of the user and the user data after the desensitization treatment need to adopt certain encryption measures to carry out encryption storage and encryption transmission during storage and transmission.

In addition, the terminal device may be further configured with an audio playing device, a video playing device, a graphic display device or the like for playing the target popularization data.

As an alternative, when the user interacts with the terminal device, the user may also use a paper strip printed with a two-dimensional code or a bar code representing the unique identifier such as the user identity or payment information to read with the reader of the terminal device, so as to transmit the user identifier to the terminal device.

In practical application, when the feature extractor is used to extract features of the above-mentioned various types of data, an appropriate feature extractor can be selected for each data according to the characteristics of different data types. For example, when extracting the flow characteristics of the device, a graph rolling network can be selected; when user feature extraction is performed, a fully connected neural network can be utilized; the graph roll-up network may be utilized in performing the extraction of the promotional features. The network type of the feature extractor can be changed or adjusted according to the requirement. This is by way of illustration only and is not to be construed as limiting the technical solutions of the present application.

The device flow feature described herein may use a certain output sub-graph in the graph convolution network as the device flow feature, so that the workload of computing the feature can be effectively reduced. And the working efficiency is improved. The specific calculation process will be specifically illustrated in the following examples.

After the features (the device flow feature, the promotion data feature and the user feature) respectively corresponding to the various types of data (the device space-time data, the data to be promoted and the user data) are obtained in the above manner, the features are further required to be fused. During fusion processing, due to different sample sizes and different feature intensities of different types of features, the dominant effect of certain features can be excessively strong, and some feature effects are not reflected, so that the obtained result is similar to the result of single-mode feature dominant. Therefore, when fusion is performed, corresponding weight coefficients are determined according to the feature importance degree. Further, the target promotion data is determined from the plurality of data to be promoted by integrating the plurality of special types. The target promotional data is determined for a first user currently interacting with the terminal device. Through the scheme, in the offline popularization scene, the accurate popularization aiming at the mobile user can be realized, and thus, a higher popularization feedback effect (such as a higher popularization click rate) can be obtained.

In one or more embodiments of the present application, obtaining user data of a plurality of different users interacting with the terminal device includes: when the first user interacts with the terminal equipment, receiving interaction request information provided by the terminal equipment and/or user portable equipment; the interaction request information comprises: the terminal device comprises a first user identifier and a terminal device identifier. The first user data is determined based on the first user identification. And determining second user data of a second user interacted with the terminal equipment according to the terminal equipment identifier, wherein the second user is a plurality of historical users interacted with the terminal equipment.

The first user is here referred to as the user who is currently interacting with the terminal device, and the second user is a history of users who have interacted with the same terminal device. When the first user and the terminal equipment interact, the first user always selectively initiates an interaction request according to own needs, and the first user and the terminal equipment are at the same place, in other words, have the same position information or position data. When the first user initiates the interaction request, a connection relationship is established between the first user and the server through the terminal device or the user portable device, and relevant data transmission is performed for the interaction, for example, the terminal device provides the user identification of the first user and the self device identification of the first user currently interacting with the first user to the server. Or providing the user identification of the first user and the equipment identification of the terminal equipment currently executing the interaction task to the server side through the user portable equipment. In general, the connection is established with the server side by the terminal device preferentially and the data interaction is performed, and when the terminal device cannot be connected with the server side, the user portable device can establish the connection with the server side and perform the related data interaction.

The server side stores user data of a first user currently interacting with the terminal equipment, user data of a second user interacted with the terminal equipment before the current time, user data generated by the interaction of the second user and the server, and related data of the terminal equipment. When the user identification and the user data are stored, the association relationship between the first user identification and the user data, between the terminal equipment identification and the equipment space-time data and between the user identification and the user data which have interacted with the terminal equipment respectively can be established in the form of a table or key value pair, so that all the associated user data, equipment space-time data and the like can be found according to the user identification and the equipment identification.

The server end not only stores the data, but also can determine target popularization data by using a preset algorithm, does not need to store and calculate the terminal equipment, and can effectively reduce the workload of the terminal equipment and reduce the configuration requirement on the terminal equipment. The specific process of determining the target promotion data by the server side through the preset algorithm will be specifically illustrated in the following embodiments.

For example, before the user finishes shopping from a supermarket and arrives at the self-service checkout equipment (corresponding to the terminal equipment), the user opens the two-dimensional code displayed by the self-service checkout equipment of the specified client-side code scanning in the mobile phone (corresponding to the portable equipment of the user), and after the code scanning is successful, the data interaction between the terminal equipment and the mobile phone of the user is completed, that is, the first user identification of the mobile phone end is sent to the self-service checkout equipment and/or the terminal equipment identification of the self-service checkout equipment is sent to the mobile phone. Furthermore, the mobile phone or the self-checkout device can send the first user identifier and the terminal device identifier to the server side.

In one or more embodiments of the present application, acquiring device space-time data and data to be promoted related to a terminal device includes: based on the terminal equipment identification, determining the equipment space-time data and a plurality of prestored data to be promoted; wherein the device spatiotemporal data comprises: and the equipment space position and the time data of the interaction between the first user or the second user and the terminal equipment.

As can be seen from the foregoing, both user data and device-related data are stored at the server side. After the server side obtains the first user identifier and the terminal equipment identifier through the steps, the terminal equipment identifier can be utilized to search the equipment space-time data and the data to be promoted which are stored before. The data to be promoted is prefabricated, and can be data in the forms of pictures, texts, videos and the like. After the terminal equipment identifier is obtained, the previously stored equipment space-time data and the time data which is interacted with the equipment can be accurately searched. And the terminal equipment is not required to be managed, so that the burden of the terminal equipment is reduced. Meanwhile, through storing the various data, the attention points of the user can be conveniently mastered, so that the popularization data can be accurately determined.

For example, the terminal equipment is self-service checkout equipment, a terminal equipment identifier ZZJZ-1 is stored in a server, the corresponding equipment position is ABC supermarket, and the time data is 3 pm. The data to be promoted is advertisements related to supermarket commodities, such as fresh promotion posters, drink promotion advertisements and the like.

For another example, the terminal device is beverage self-service vending device, the terminal device identifier is ZZSM-1 stored in the server, the corresponding device position is BCD market, and the time is 11 noon. The data to be promoted is an advertisement related to each merchant of the market, such as a certain chafing dish advertisement, a certain daily material advertisement, a certain clothing advertisement, etc. Since lunch time is near, food-related advertisements can be selected as target recommendation data in a targeted manner.

In one or more embodiments of the present application, the extraction method of the flow characteristics of the device includes: and extracting the characteristics of the equipment space-time data and the second user data by using a graph rolling network, and representing the obtained low-level subgraph of the graph rolling network as equipment flow characteristics.

In practical application, when generating the flow characteristics of the device, not only the device is the space-time data of the device, but also the second user data of the second user having interacted with the terminal device in the history period are considered, and before using the second user data, sensitive data including basic identity information, consumption data and the like of the second user need to be desensitized or statistically processed. The second user data also includes feedback data for historical targeted promotional data for the second user that the second user provided for the terminal device, such as whether the second user clicked or selected advertising content for which the terminal device promoted. Thereby enabling more accurate flow characteristics of the device.

For example, by using a graph construction technology, graph data is obtained based on equipment space-time data and second user data, and then convolution processing is performed on the graph data to obtain subgraphs of different orders. In order to reduce the computational effort and enable a relatively accurate representation of the context, as an alternative, a second order subgraph output may be selected as the device flow feature.

In one or more embodiments of the present application, the extraction manner of the promotion data feature includes: performing feature extraction on the data to be promoted by using a pre-trained convolutional neural network to obtain the promotion features; the convolutional neural network is obtained by training a shallow layer in the convolutional neural network model based on a popularization data training sample.

In practical applications, convolutional neural networks, such as VGG-16, may be utilized when processing data to be generalized. Because the number of layers in the convolutional neural network is relatively large, after training is completed by using the basic samples, in practical application, fine adjustment is also required according to the type of the actually processed image. To alleviate the trimming effort, the deep layers in the convolutional network may be fixed, with only the shallow layers being trimmed. It should be noted that, which shallow layers are fine-tuned and to what extent, it is necessary to determine according to the actual needs of the user. In order to reduce the workload of adjustment and improve the adjustment efficiency, a small amount of shallow layers can be selected for fine adjustment. If the trimming effect is not good at this time, the number of shallow layers to be trimmed can be increased. Through the mode, the workload of optimizing the convolutional neural network is reduced as much as possible while the requirements on the feature extraction of the popularization data are met.

In one or more embodiments of the present application, multimodal fusion of the device flow characteristics, the promotional data characteristics, and the user characteristics includes:

based on an attention mechanism, determining a first weight corresponding to the equipment flow feature, a second weight corresponding to the promotion data feature and a third weight corresponding to the user feature;

respectively carrying out dot product processing on the equipment flow characteristics, the first weight, the promotion data characteristics, the second weight and the user characteristics and the third weight to obtain a plurality of modal characteristics;

and carrying out fusion processing on the plurality of modal features by using a preset model.

Specifically, fig. 3 is a schematic diagram illustrating feature fusion according to an embodiment of the present application. In order to alleviate the problem of inconsistency among various types of data, features can be extracted from each mode respectively, and then fusion is performed at a feature level, namely feature fusion. Since deep learning involves learning a specific representation of a feature from the original data, sometimes resulting in data fusion before the feature is not extracted, the fusion of the data plane and the feature plane is referred to as early fusion. As can be seen from fig. 3, the device flow characteristics, the user characteristics, and the promotion data characteristics are respectively calculated by dot product, so as to obtain the fusion characteristics.

In the implementation process of feature fusion, firstly, each input mode feature is extracted, then the extracted features are combined into fusion features, the fusion features are input into a recommendation model as input data, and a prediction result is output. The fused features generated by converting and scaling the modal features generally have higher dimensionality, and the fused features can be subjected to dimension reduction processing by using Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). There are several fusion modes, common ones are multiplication or addition of identical position elements for each modality feature, construction of encoder-decoder structure and use of long short-term memory (LSTM).

The weights referred to herein may be calculated based on the attention i on mechanism. Fig. 4 is a schematic flow chart of weight-based computation fusion according to an embodiment of the present application. Suppose now that a set of inputs is to be entered

H= [ H1, H2, H3,., hn ] (e.g., user profile or device flow profile) calculates important content using the content mechanism, where a query vector q is often needed, and then the correlation between the query vector q and each input H i is calculated by a scoring function, yielding a score. These scores are then normalized using softmax, the normalized result being the attention distribution a= [ a1, a2, a3, ], of the query vector q over the respective inputs H i, where each term value corresponds one-to-one to the original input h= [ H1, H2, H3, ], hn. Taking a i as an example, the correlation calculation formula is as follows:

a i＝softmax(s(h i,q))＝exp(s(h i,q))∑j＝1nexp(s(hj,q))

Wherein the scoring function may be calculated using several ways:

additive model s (h, q) =vttanh (wh+uq)

Dot product model s (h, q) = hTq

Scaling the dot product model s (h, q) =htqd

Bilinear model s (h, q) =htqq

The parameters in the above formulas, and W, U and v, are each a matrix or vector of parameters that can be learned, and D is the dimension of the input vector. The specific user can select an appropriate calculation mode according to the needs.

Through the scheme, when fusion processing is carried out on a plurality of different types of features, in order to balance the weight of each feature well, the correlation coefficient and the fusion weight are calculated based on the attention mechanism, so that the feature weights are self-adaptively adjusted, the fused features obtained through the self-adaptive adjustment are further utilized to select popularization information, the popularization requirements are met, meanwhile, the validity of multi-modal feature fusion is ensured, the bad influence of multi-modal fusion is eliminated, for example, the multi-modal competition is caused, the model effect of the multi-modal is called as single-modal, and the accuracy of an evaluation result is seriously influenced.

In one or more embodiments of the present application, after the device flow feature, the promotion data feature, and the user feature are multimodal fused and input into a recommendation model, further comprising:

Inputting the fused multi-modal characteristics into the recommendation model;

determining clicking rates corresponding to the data to be promoted respectively;

target promotion data aiming at the terminal equipment are determined from a plurality of data to be promoted according to the click rate;

and sending the target promotion data to the terminal equipment which interacts with the first user and/or the user portable equipment of the first user.

After the fusion features are obtained in the manner described above, a recommendation model (e.g., a click-Through-Rate model (CTR)) is used to calculate the click Rate of each data to be promoted. After the click rate is calculated, the target popularization data suitable for the first user are further screened out according to the click rate.

It should be noted that, the click feedback result of the target promotion data promoted by the terminal device needs to be sent to the server side and stored by the server side, so as to be used as a part of the user data of the first user, so as to be used as the user data of the subsequent historical user, and be used as the basic data used in the next promotion data. Through the scheme, popularization data can be provided for users more accurately.

For example, the terminal device is a beverage self-service vending device, and the self-service vending device can perform data interaction with the server side. The server side stores a terminal equipment identifier ZZSM-1, the corresponding equipment position is a BCD market, and the time is 11 noon. The data to be promoted is an advertisement related to each merchant of the market, such as a certain chafing dish advertisement, a certain daily material advertisement, a certain clothing advertisement, etc. Since lunch time is near, food-related advertisements can be selected as target recommendation data in a targeted manner. Because the first user history food consumption record recorded by the server side is spicy, and the fire pot consumption record is recorded for a plurality of times, the target promotion data is determined to be a certain fire pot advertisement by utilizing the various types of data, the feature extractor and the CTR model.

Based on the same thought, the embodiment of the application also provides another data processing method. The method may be performed by a terminal device, which is a device providing a service to a non-stationary user. Fig. 5 is a flow chart of another data processing method according to an embodiment of the present application. As can be seen from fig. 5, the method specifically comprises the steps of:

501: and responding to the interaction request of the first user through the client, and acquiring the first user identification.

502: the first user identification and the equipment identification of the terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to extract features of the equipment space-time data, the data to be promoted and the user data, and equipment flow features, promotion data features and user features represented by a designated sub-graph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data.

503: and receiving target promotion data for the terminal equipment, which are determined by the server side from a plurality of data to be promoted by utilizing the recommendation model.

The method corresponds to the system shown in fig. 1a of the drawings of the specification. The terminal equipment can establish connection with the server and conduct data interaction. The user portable equipment does not need to participate in a popularization data transmission task, the terminal equipment can send the acquired first user identification and terminal equipment identification to the server side, and the server side searches various types of basic data based on the acquired first user identification and terminal equipment identification, performs feature extraction by using feature extractors aiming at different types of data and determines target popularization data which needs to be recommended to the first user by using a recommendation model. The target promotion data are sent to the terminal equipment, so that the first user can know the content of the target promotion data through the terminal equipment and decide whether to click the data according to the needs of the first user.

Based on the same thought, the embodiment of the application also provides a data processing method. The method is applied to the user portable equipment provided with the client. Fig. 6 is a flow chart of another data processing method according to an embodiment of the present application. As can be seen from fig. 6, the method specifically comprises the steps of:

601: and after the interaction request is sent to the terminal equipment, receiving feedback information carrying the equipment identifier.

602: the method comprises the steps that a first user identification of a first user in current interaction with terminal equipment and equipment identification of the terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to extract features of the equipment space-time data, the data to be promoted and the user data, and equipment flow features, promotion data features and user features represented by a designated sub-graph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data.

603: and receiving target promotion data for the terminal equipment, which are determined by the server side from a plurality of data to be promoted by utilizing the recommendation model.

This method corresponds to the system shown in fig. 1b of the drawing of the specification. The terminal device cannot establish a connection with the server and perform data interaction. The user portable equipment is required to participate in the popularization data transmission task, the obtained first user identification and terminal equipment identification can be sent to the server side by the user portable equipment, and then the server side searches various types of basic data based on the obtained first user identification and terminal equipment identification, and utilizes a feature extractor aiming at different types of data to conduct feature extraction and utilizes a recommendation model to determine target popularization data which needs to be recommended to the first user. The target promotion data are sent to the user portable equipment, so that the first user can know the content of the target promotion data through the user portable equipment, and whether to click the data is determined according to the needs of the first user.

Based on the same thought, the embodiment also provides a data processing device. Fig. 7 is a schematic structural diagram of a data processing apparatus according to an embodiment of the present application. As can be seen from fig. 7, the device comprises:

and the acquiring module 71 is configured to acquire device space-time data related to a terminal device, data to be promoted, and user data of a plurality of different users interacting with the terminal device, where the user data is desensitized data or statistical data, and the terminal device is a device that provides services to a non-fixed user.

And the feature extraction module 72 is configured to perform feature extraction on the device space-time data, the data to be promoted and the user data by using a feature extractor, so as to obtain a device flow feature, a promotion data feature and a user feature represented by a designated sub-graph.

And the feature fusion module 73 is configured to perform multi-mode fusion on the device flow feature, the promotion data feature, and the user feature, and input the multi-mode fusion into a recommendation model, so as to determine target promotion data for a current user in interaction with the terminal device from a plurality of data to be promoted by using the recommendation model.

An obtaining module 71, configured to receive interaction request information provided by the terminal device and/or a portable device of the user when the first user is interacting with the terminal device; the interaction request information comprises: a first user identifier and a terminal device identifier;

determining the first user data based on the first user identification;

and determining second user data of a second user interacted with the terminal equipment according to the terminal equipment identifier, wherein the second user is a plurality of historical users.

An obtaining module 71, configured to determine, based on the terminal device identifier, the device spatiotemporal data, and a plurality of prestored data to be promoted; wherein the device spatiotemporal data comprises: and the equipment space position and the time data of the interaction between the first user or the second user and the terminal equipment.

The feature extraction module 72 is configured to perform feature extraction on the device spatiotemporal data and the second user data by using a graph rolling network, and represent a low-level sub-graph of the obtained graph rolling network as a device flow feature.

The feature extraction module 72 is configured to perform feature extraction on the data to be promoted by using a pretrained convolutional neural network, so as to obtain the promotion feature; the convolutional neural network is obtained by training a shallow layer in the convolutional neural network model based on a popularization data training sample.

A feature fusion module 73, configured to determine, based on an attention mechanism, a first weight corresponding to the device flow feature, a second weight corresponding to the promotion data feature, and a third weight corresponding to the user feature;

A promotion module 74 is further included for inputting the multi-modal feature after fusion into the recommendation model;

Based on the same idea, the own embodiment also provides another data processing device. Fig. 8 is a schematic structural diagram of another data processing apparatus according to an embodiment of the present application. As can be seen from fig. 8, the device comprises:

the obtaining module 81 is configured to obtain the first user identifier in response to an interaction request of the first user through the client.

A sending module 82, configured to send the first user identifier and the device identifier of the terminal device to a server, so that the server performs feature extraction on the device space-time data, the data to be promoted, and the user data by using a feature extractor, to obtain a device flow feature, a promotion data feature, and a user feature that are represented by a specified sub-graph; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data.

And the receiving module 83 is configured to receive target promotion data for the terminal device, which is determined by the server side from a plurality of data to be promoted by using the recommendation model.

Based on the same idea, the own embodiment also provides a further data processing device. Fig. 9 is a schematic structural diagram of still another data processing apparatus according to an embodiment of the present application. As can be seen from fig. 9, the device comprises:

and the receiving module 91 is configured to receive feedback information carrying the device identifier after sending the interaction request to the terminal device.

A sending module 92, configured to send a first user identifier of a first user currently interacting with the terminal device and a device identifier of the terminal device to a server, so that the server performs feature extraction on the device space-time data, the data to be promoted, and the user data by using a feature extractor, to obtain a device flow feature, a promotion data feature, and a user feature that are represented by a specified sub-graph; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data.

The receiving module 91 is configured to receive target promotion data for the terminal device, which is determined by the server side from a plurality of data to be promoted by using the recommendation model.

For ease of understanding, the following is illustrated in connection with an actual scenario.

User a just sees the movie at the mall, and purchases the drinking water on the vending machine through scanning the code, which is a good marketing opportunity for the merchant at this time, but how to realize accurate popularization marketing needs to be considered. However, since the user a is a new user with respect to the vending machine, the conventional marketing system cannot acquire the historical behavior data, so that accurate marketing is performed, the marketing content is not the content currently required by the user a, and the opportunity is finally missed. Aiming at the phenomenon, the commonality can be extracted through the user data, the equipment space-time data and the data to be promoted stored at the server side for providing other services for a plurality of users, so that the user can be known to possibly go to a restaurant to eat next, and the related content of the food is marketed, so that the terminal equipment realizes the accurate promotion and marketing aiming at the new user of the first interaction.

Furthermore, in the off-line retail scene, data such as user behavior, to-be-promoted data (e.g., video advertisements, picture advertisements), equipment space-time attribute and the like need to be combined to comprehensively learn a multi-mode recommendation model. In this case, more accurate popularization results are expected while more different types of data are fused. However, in the practical application process, a single mode (such as a user feature) is easy to lead the whole feature composition, other features (such as a popularization data feature and the like) cannot play a role, and the whole multi-mode system is rare. Therefore, the multi-mode feature fusion module realized by adopting the attention mechanism can better fuse the multi-mode features, and simultaneously avoid that a single mode leads the whole system.

An embodiment of the application also provides a cloud server. The cloud server is a master node electronic device in the computing unit. Fig. 10 is a schematic structural diagram of a cloud server according to an embodiment of the present application. The electronic device includes a memory 1001, a processor 1002, and a communication component 1003; wherein,,

the memory 1001 is configured to store a program;

the processor 1002, coupled to the memory, is configured to execute the program stored in the memory, for:

Optionally, the processor 1002 is configured to receive interaction request information provided by the terminal device and/or the user portable device when the first user is interacting with the terminal device; the interaction request information comprises: a first user identifier and a terminal device identifier;

determining the first user data based on the first user identification;

Optionally, the processor 1002 is configured to determine, based on the terminal device identifier, the device spatiotemporal data, and a plurality of pre-stored data to be promoted; wherein the device spatiotemporal data comprises: and the equipment space position and the time data of the interaction between the first user or the second user and the terminal equipment.

Optionally, the processor 1002 is configured to perform feature extraction on the device spatiotemporal data and the second user data by using a graph rolling network, and represent a resulting low-level sub-graph of the graph rolling network as a device flow feature.

Optionally, the processor 1002 is configured to perform feature extraction on the data to be promoted by using a pre-trained convolutional neural network, so as to obtain the promotion feature; the convolutional neural network is obtained by training a shallow layer in the convolutional neural network model based on a popularization data training sample.

Optionally, the processor 1002 is configured to determine, based on an attention mechanism, a first weight corresponding to the device flow feature, a second weight corresponding to the promotion data feature, and a third weight corresponding to the user feature;

Optionally, the processor 1002 is configured to input the multimodal features after fusion into the recommendation model;

The memory 1001 described above may be configured to store various other data to support operations on an electronic device. Examples of such data include instructions for any application or method operating on an electronic device. The memory may be implemented by any type of volatile or nonvolatile memory device or combination thereof, such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.

Further, the processor 1002 in this embodiment may specifically be: and the programmable exchange processing chip is provided with a data copying engine which can copy the received data.

The processor 1002 may perform other functions in addition to the above functions when executing programs in a memory, and specific reference may be made to the foregoing descriptions of embodiments. Further, as shown in fig. 10, the electronic device further includes: power supply assembly 1004, and the like.

Embodiments also provide a non-transitory machine-readable storage medium having executable code stored thereon, which when executed by a processor of an electronic device, causes the processor to perform the method described in the corresponding embodiment of fig. 2.

One embodiment of the present application also provides another electronic device. The electronic device is a master node electronic device in the computing unit. The electronic device is a device that provides services to a non-stationary user. Fig. 11 is a schematic structural diagram of another electronic device according to an embodiment of the present application. The electronic device comprises a memory 1101, a processor 1102 and a communication component 1103; wherein,,

The memory 1101 is configured to store a program;

the processor 1102 is coupled to the memory for executing the program stored in the memory for:

the first user identification and the equipment identification of the terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to extract features of the equipment space-time data, the data to be promoted and the user data, and equipment flow features, promotion data features and user features represented by a designated sub-graph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data;

The memory 1101 described above may be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on an electronic device. The memory may be implemented by any type of volatile or nonvolatile memory device or combination thereof, such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.

Further, the processor 1102 in this embodiment may specifically be: and the programmable exchange processing chip is provided with a data copying engine which can copy the received data.

The processor 1102 may perform other functions in addition to the above functions when executing programs in memory, and specific reference may be made to the foregoing descriptions of embodiments. Further, as shown in fig. 11, the electronic device further includes: power supply component 1104, and the like.

Embodiments also provide a non-transitory machine-readable storage medium having executable code stored thereon, which when executed by a processor of an electronic device, causes the processor to perform the method described in the corresponding embodiment of fig. 5.

One embodiment of the present application also provides yet another electronic device. The electronic device is a master node electronic device in the computing unit. Fig. 12 is a schematic structural diagram of still another electronic device according to an embodiment of the present application. The electronic device comprises a memory 1201, a processor 1202 and a communication component 1203; wherein,,

the memory 1201 is used for storing a program;

The processor 1202, coupled to the memory, is configured to execute the program stored in the memory for:

the method comprises the steps that a first user identification of a first user in current interaction with terminal equipment and equipment identification of the terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to extract features of the equipment space-time data, the data to be promoted and the user data, and equipment flow features, promotion data features and user features represented by a designated sub-graph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data;

The memory 1201 may be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on an electronic device. The memory may be implemented by any type of volatile or nonvolatile memory device or combination thereof, such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.

Further, the processor 1202 in this embodiment may specifically be: and the programmable exchange processing chip is provided with a data copying engine which can copy the received data.

The processor 1202 may perform other functions in addition to the above functions when executing programs in memory, as described in detail in the foregoing embodiments. Further, as shown in fig. 12, the electronic device further includes: power supply assembly 1204, and the like.

Embodiments also provide a non-transitory machine-readable storage medium having executable code stored thereon, which when executed by a processor of an electronic device, causes the processor to perform the method described in the corresponding embodiment of fig. 6.

Based on the above embodiment, in the online scene, some terminal devices are not devices that provide services to unique users, but devices that provide services to non-stationary users. The server side can directly or indirectly provide promotion data for the terminal equipment, so that the promotion data better meets the requirement of a first user currently interacting with the terminal equipment, the server side obtains more comprehensive data, the data comprise equipment space-time data, data to be promoted and user data of a plurality of different users interacting with the terminal equipment (including the user data currently interacting and interacted), and the user data is desensitized data or statistical data. Further, the feature extraction is performed by a feature extractor, and the corresponding features are obtained as follows: device flow characteristics, promotional data characteristics, and user characteristics represented by the specified sub-graph. And further fusing the characteristics corresponding to different data types, and determining target popularization data suitable for the user interacting with the terminal equipment at present through a recommendation model. The users with the interaction behavior with the terminal equipment often have the same or similar characteristics, so that the appointed sub-graph representation can be selected to serve as the equipment flow characteristic under the condition of considering the calculation efficiency and the characteristic extraction effect, the common characteristics of the users with the interaction behavior with the terminal equipment can be comprehensively represented, and the calculation amount can be effectively reduced. In addition, user data experienced by the terminal equipment with larger user mobility is collected, the terminal flow characteristics of the terminal equipment are accurately extracted by utilizing the corresponding characteristic extractor, the user characteristics of the current user can be further integrated, the promotion data characteristics of the data to be promoted and the equipment flow characteristics comprehensively evaluate the target promotion data recommended to the user through the terminal equipment, the accuracy of the terminal equipment in data promotion can be effectively improved, and the user can obtain the promotion data required by the user on strange terminal equipment.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and provide corresponding operation entries for the user to select authorization or rejection.

The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims

1. A data processing method, applied to a server, the method comprising:

2. The method of claim 1, wherein obtaining user data for a plurality of different users interacting with the terminal device comprises:

when a first user interacts with the terminal equipment, receiving interaction request information provided by the terminal equipment and/or user portable equipment; the interaction request information comprises: a first user identifier and a terminal device identifier;

determining the first user data based on the first user identification;

3. The method of claim 2, wherein obtaining device space-time data and data to be promoted related to the terminal device comprises:

based on the terminal equipment identification, determining the equipment space-time data and a plurality of prestored data to be promoted; wherein the device spatiotemporal data comprises: and the equipment space position and the time data of the interaction between the first user or the second user and the terminal equipment.

4. The method according to claim 2, wherein the extraction of the flow characteristics of the device comprises:

and extracting the characteristics of the equipment space-time data and the second user data by using a graph rolling network, and representing the obtained low-level subgraph of the graph rolling network as equipment flow characteristics.

5. The method of claim 1, wherein the extraction of the promotional data features comprises:

performing feature extraction on the data to be promoted by using a pre-trained convolutional neural network to obtain the promotion features; the convolutional neural network is obtained by training a shallow layer in the convolutional neural network model based on a popularization data training sample.

6. The method of claim 2, wherein multimodal fusion of the device flow characteristics, the promotional data characteristics, and the user characteristics comprises:

7. The method of claim 6, wherein after multimodal fusion of the device flow characteristics, the promotional data characteristics, and the user characteristics and input into a recommendation model, further comprising:

inputting the fused multi-modal characteristics into the recommendation model;

8. A data processing method applied to a terminal device, the terminal device being a device that provides a service to a non-stationary user, the method comprising:

9. A data processing method applied to a user portable device mounted with a client, the method comprising:

the method comprises the steps that a first user identifier of a first user in current interaction with terminal equipment and the equipment identifier of the terminal equipment are sent to a server side, so that the server side utilizes a feature extractor to extract features of equipment space-time data, data to be promoted and user data, and equipment flow features, promotion data features and user features represented by a designated sub-graph are obtained; carrying out multi-mode fusion on the equipment flow characteristics, the promotion data characteristics and the user characteristics and inputting the multi-mode fusion into a recommendation model; the user data is desensitization data or statistical data;

10. A data processing system, the system comprising:

A server side for executing the method of any one of claims 1 to 7;

terminal device for performing the method of claim 8.

11. A data processing system, the system comprising:

a server side for executing the method of any one of claims 1 to 7;

a user portable device for performing the method of claim 9.

12. A cloud server comprises a memory and a processor; wherein,,

the memory is used for storing programs;

the processor, coupled to the memory, for executing the program stored in the memory for implementing the method of any of the preceding claims 1 to 7.

13. A non-transitory machine-readable storage medium having stored thereon executable code which, when executed by a processor of an electronic device, causes the processor to perform the method of any of claims 1 to 7, or the method of claim 8, or the method of claim 9.