Service recommendation method

Abstract

The invention discloses a service recommendation method, which is used for improving the effectiveness of recommending service content to a user, and the scheme provided by the invention comprises the following steps: obtaining position information reported by a user terminal, the user terminal being preset with a super SIM (Subscriber Identity Module) card; if the position information represents that the user terminal is located in a preset electronic fence, the user terminal is indicated to display a super SIM message, and the super SIM message comprises information of an optional service item corresponding to the preset electronic fence; based on an interactive operation on the selectable service item received by the user terminal, executing identity verification through the super SIM card; based on the identity verification result, the user terminal is indicated to display service content matched with the identity information of the super SIM card through a 5G message, and the service content is the service content predicted by a pre-trained service content recommendation model based on the identity information.

Description

Technical Field

[0001] The present application relates to the field of communications, and in particular to a service recommendation method. Background Art

[0002] In the field of communication services, the types of communication services are becoming more and more diversified. Users can see a lot of information through user terminals, but the service information that matches the user is difficult to effectively reach the user.

[0003] If users actively search for services, they can use the required service functions, but they need to perform tedious operations. If services are recommended to users, so that users passively view service information, there may be a situation where the recommended services do not match the users. In this case, not only can users not use the required services conveniently, but they may also be disturbed by unnecessary information.

[0004] How to improve the effectiveness of recommending service content to users is the technical problem to be solved by this application. Summary of the invention

[0005] The purpose of the embodiment of the present application is to provide a service recommendation method to improve the effectiveness of recommending service content to users.

[0006] In a first aspect, a service recommendation method is provided, comprising: Acquiring location information reported by a user terminal, wherein the user terminal is pre-installed with a super SIM card; If the location information indicates that the user terminal is located within a preset electronic fence, instructing the user terminal to display a super SIM message, where the super SIM message includes information about optional service items corresponding to the preset electronic fence; Based on the interactive operation on the optional service item received by the user terminal, performing identity verification through the super SIM card; Based on the identity verification result, the user terminal is instructed to display service content matching the identity information of the super SIM card through a 5G message, wherein the service content is the service content predicted by a pre-trained service content recommendation model based on the identity information.

[0007] In a second aspect, a service recommendation device is provided, comprising: An acquisition module, for acquiring location information reported by a user terminal, wherein the user terminal is pre-installed with a super SIM card; an indication module, which indicates that the user terminal is located within a preset electronic fence if the location information indicates that the user terminal is located within the preset electronic fence, wherein the super SIM message includes information about optional service items corresponding to the preset electronic fence; A verification module, performing identity verification through the super SIM card based on the interactive operation on the optional service item received by the user terminal; A display module instructs the user terminal to display service content matching the identity information of the super SIM card through a 5G message based on the identity verification result, wherein the service content is the service content predicted by a pre-trained service content recommendation model based on the identity information.

[0008] According to a third aspect, an electronic device is provided. The electronic device comprises a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps of the method according to the first aspect are implemented.

[0009] According to a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method according to the first aspect are implemented.

[0010] In a fifth aspect, a computer program product is provided, the computer program product comprising a non-transitory computer-readable storage medium storing a computer program, the computer program being operable to cause a computer to execute part or all of the steps of the method of the first aspect.

[0011] In an embodiment of the present application, first, the location information reported by the user terminal is obtained, and the user terminal is pre-installed with a super SIM card; if the location information indicates that the user terminal is located within a preset electronic fence, the user terminal is instructed to display a super SIM message, wherein the super SIM message includes information about optional service items corresponding to the preset electronic fence; then, based on the interactive operation of the optional service items received by the user terminal, identity verification is performed through the super SIM card; finally, based on the identity verification result, the user terminal is instructed to display service content matching the identity information of the super SIM card through a 5G message, wherein the service content is the service content predicted by the pre-trained service content recommendation model based on the identity information. Through this solution, the information of the optional service items is accurately provided to the user based on the location information reported by the user terminal, and then a reliable and authentic identity verification is performed through the super SIM card. Subsequently, the display is performed through the 5G message, which can improve the richness of the message content, realize multi-form display of the service content matching the identity information, and optimize the readability of the service content. Among them, the service content is predicted by the pre-trained service content recommendation model based on the identity information, which can improve the matching degree between the displayed service content and the identity information of the super SIM card, and enhance the attractiveness of the recommended service to the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1a This is one of the flowcharts of a service recommendation method in one embodiment of the present application; Figure 1b This is a system structure and function diagram of a service platform for executing a service recommendation method according to an embodiment of the present application; Figure 1c This is a schematic diagram of a scenario application for executing a service recommendation method according to an embodiment of the present application; Figure 2 This is a second flow chart of a service recommendation method according to an embodiment of the present application; Figure 3a This is a flowchart of a service recommendation method according to an embodiment of the present application; Figure 3b This is a schematic diagram of a model training process of a service recommendation method in an embodiment of the present application; Figure 4 This is a fourth flowchart of a service recommendation method according to an embodiment of the present application; Figure 5 This is a flowchart of a service recommendation method according to an embodiment of the present application; Figure 6 This is a flowchart of a service recommendation method according to an embodiment of the present application; Figure 7 This is the seventh flowchart of a service recommendation method in an embodiment of the present application; Figure 8a This is a flowchart of a service recommendation method according to an embodiment of the present application; Figure 8b It is a flowchart of a merchant side of a service recommendation method in an embodiment of the present application; Figure 9a This is a ninth flowchart of a service recommendation method according to an embodiment of the present application; Figure 9b This is a flowchart of a user side of a service recommendation method according to an embodiment of the present application; Fig.10 This is a flowchart of a service recommendation method according to an embodiment of the present application; Fig.11 It is a structural diagram of a service recommendation device according to an embodiment of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application. The numbering of the drawings in this application is only used to distinguish the various steps in the scheme, and is not used to limit the execution order of the various steps. The specific execution order is subject to the description in the specification.

[0014] In the service recommendation application scenario, relevant information about service recommendations can be sent to users through applications (Application, APP), Web (World Wide Web, Global Wide Area Network) pages, etc. Specifically, after obtaining the geographical location or IP (Internet Protocol) address permission of the user terminal based on the LBS (Location Based Services, LBS) recommendation system, users can be reached through in-site messages, PUSH (application message) push, in-terminal streaming media advertising, etc.

[0015] The drawbacks of these service recommendation methods are that, on the one hand, push messages may be limited by message permissions and cannot be effectively displayed to users in real time. On the other hand, it is difficult to make targeted service recommendations to users. For example, to obtain the user's identity information and then determine the user's needs, the user is often required to perform cumbersome operations. Moreover, the display effect of messages such as pop-up windows and text messages in the background of the application is limited. As a result, the click-through rate of actual service recommendations is low.

[0016] In order to solve the problems existing in the related art, the embodiment of the present application provides a service recommendation method. This solution can be executed by the service platform, such as Figure 1a As shown, the following steps are included: S11: Acquire location information reported by a user terminal, where the user terminal is pre-installed with a super SIM card.

[0017] The above-mentioned location information can be information reported by the user terminal to the communication base station based on the user's authorized functions, and then provided by the communication base station to the service platform. Among them, the location information reported by the user terminal can be used to provide communication services to the user terminal through the communication base station. The location information is authentic and reliable, which can help improve the quality of service recommendations.

[0018] The super SIM (Subscriber Identity Module) card can be inserted into the user terminal through a preset slot of the user terminal.

[0019] A SIM card is a smart card that can be used to store user identification data, text message data, and phone numbers. Super SIM cards have more practical functions than SIM cards, such as swiping cards through Near Field Communication (NFC) and identity verification through 5G electronic signatures. In addition, super SIM cards often have larger storage space than SIM cards and can store more information during use.

[0020] In the embodiment of the present application, a super SIM card corresponds to the identity information of a communication user, and a communication user may hold one or more super SIM cards.

[0021] S12: If the location information indicates that the user terminal is located within a preset electronic fence, instruct the user terminal to display a super SIM message, where the super SIM message includes information about optional service items corresponding to the preset electronic fence.

[0022] In the field of mobile communication technology, electronic fences can be pre-set based on real geographical locations. By setting electronic fences, different geographical areas can be divided, and customized service functions can be provided for different location areas.

[0023] In the embodiment of the present application, the preset electronic fence can be pre-set by the service provider. For example, the merchant can set a circular electronic fence with the location of the physical store as the center and a certain distance as the radius. When a user enters the area with a mobile terminal, the electronic fence can effectively identify the user terminal that can provide services.

[0024] In actual applications, different service providers can set different electronic fences according to their own needs. The electronic fences can also be customized based on time periods, such as setting one electronic fence for weekdays and another electronic fence for weekends. Setting electronic fences can facilitate service providers to flexibly provide a variety of services.

[0025] The above service provider can set the information of the corresponding optional service items for the electronic fence, and the optional service items can specifically include the service content that the service provider can provide to the users within the electronic fence. The information of the optional service items can include information in various forms such as text, pictures, and videos, and the content of the optional service can be displayed to the user through the information of the optional service items. Furthermore, the user can select the required service items through interactive operations according to his actual needs.

[0026] In an embodiment of the present application, if the location information indicates that the user terminal is located within a preset electronic fence, the user terminal is instructed to display a super SIM message. The super SIM message can be implemented based on 5G (5th Generation Mobile Communication Technology), and the super SIM message can also be called a super SIM must-reach message. In this step, the service platform issues an instruction to the super SIM card within the preset electronic fence to display the super SIM message on the plugged-in user terminal through the super SIM card.

[0027] For example, a super SIM message may include a content introduction of a recommended optional service and provide two options of confirming to accept the service and rejecting the service. Alternatively, a super SIM message may include multiple optional service items, and the user may select the required service item by checking the confirmation box. In actual applications, the optional service items may also be displayed in other forms.

[0028] S13: Based on the interactive operation on the optional service item received by the user terminal, perform identity verification through the super SIM card.

[0029] If the user performs an interactive operation on the optional service item in the super SIM message, the super SIM card is authenticated. Specifically, the super SIM card can interact with the super SIM card application to exchange identity authentication messages, and an encrypted data channel is used to transmit identity authentication related messages during the interaction process to ensure the security and effectiveness of identity authentication.

[0030] The interactive operation received by the user terminal may refer to an operation performed by the user, and the interactive operation indicates that the user has selected an accepted service item from the optional service items. In this step, after the user selects the service item and before providing the service corresponding to the service item, the identity of the user's super SIM card is verified, which can realize user identity authentication and is conducive to providing customized services to the user in the future, so as to fully meet the user's personalized service needs.

[0031] S14: Based on the identity verification result, instruct the user terminal to display service content matching the identity information of the super SIM card through a 5G message, wherein the service content is the service content predicted by a pre-trained service content recommendation model based on the identity information.

[0032] The identity verification result may indicate whether the identity information of the super SIM card is authorized to use the service item selected by the user. The identity verification result may also include information required for determining the matching service content.

[0033] For example, the pre-trained service content recommendation model may specifically be a generative pre-trained model (Pretrained Foundation Models, PFMs) based on a Transformer framework.

[0034] Optionally, when the identity verification result indicates that the identity information is authorized to use the above-mentioned service items, the identity information of the super SIM card is input into a pre-trained service content recommendation model to obtain the service content predicted by the model.

[0035] Optionally, if the identity verification result also carries other user portrait information corresponding to the identity information, the identity information and user portrait information and other information with user characteristics can be input into the above-mentioned pre-trained service content recommendation model to provide the model with sufficient user characteristics, so as to obtain service content that fits the user characteristics output by the model.

[0036] Among them, the service content is predicted by the pre-trained service content recommendation model, and the service content that meets the user's personalization can be predicted based on the user's characteristics. For example, in the scenario of product recommendation, the pre-trained model can recommend product stores and related discount information that meet the user's age group based on the user's identity information, thereby showing the user attractive service content.

[0037] In this step, the user terminal is instructed to display the service content matching the identity information of the super SIM card through 5G messages. Among them, 5G messages can be rich media messages, which can fully display information in various forms such as text, pictures, voice, and video. 5G messages can enrich the expressiveness of service content and enable users to intuitively experience service content.

[0038] In an embodiment of the present application, first, the location information reported by the user terminal is obtained, and the user terminal is pre-installed with a super SIM card; if the location information indicates that the user terminal is located within a preset electronic fence, the user terminal is instructed to display a super SIM message, wherein the super SIM message includes information about optional service items corresponding to the preset electronic fence; then, based on the interactive operation of the optional service items received by the user terminal, identity verification is performed through the super SIM card; finally, based on the identity verification result, the user terminal is instructed to display service content matching the identity information of the super SIM card through a 5G message, wherein the service content is the service content predicted by the pre-trained service content recommendation model based on the identity information. Through this solution, the information of the optional service items is accurately provided to the user based on the location information reported by the user terminal, and then a reliable and authentic identity verification is performed through the super SIM card. Subsequently, the display is performed through the 5G message, which can improve the richness of the message content, realize multi-form display of the service content matching the identity information, and optimize the readability of the service content. Among them, the service content is predicted by the pre-trained service content recommendation model based on the identity information, which can improve the matching degree between the displayed service content and the identity information of the super SIM card, optimize the quality of the recommended service content, and enhance the attractiveness of the recommended service to users.

[0039] Next, this solution is further explained in conjunction with an application scenario.

[0040] The service recommendation method provided in the embodiment of the present application can be used to recommend a merchant's services to users, and the service platform that executes the service recommendation method may include a communication base station. Figure 1b A system structure function diagram of a service platform for executing a service recommendation method is shown.

[0041] First, obtain the required data to provide a data basis for the subsequent execution of service recommendations. Among them, the data basis may include base station data, merchant and customer behavior preferences, merchant and customer static attributes, model generation results of pre-trained models for customer predictions, manual marking results, etc. Among them, base station data may specifically include base station geographic location data, which can be used to implement the electronic fence function. Merchant and customer static data can be obtained through the merchant and customer information library. Merchant and customer behavior preferences or other non-static attributes can be obtained through merchant and user behavior data. For example, behavior data may include relevant data on setting parameters, sliding clicks, uplink messages, ordering payments, feedback, and other behaviors.

[0042] The service platform includes a card management platform, which can be a super SIM card management platform. It can implement identity authentication and chip-level security protection based on the super SIM card. It can interact with the user terminal through the application installed on the USIM card to execute commands, so that the must-reach message window pops up on the user terminal to achieve effective delivery of the super SIM message.

[0043] The service platform includes a 5G messaging platform, which can specifically involve subscription account services and industry account services. For merchants, this platform allows them to flexibly implement service recommendation settings based on their own services. This may include selecting message templates, push strategies, reach times, etc., so as to achieve personalized configuration of merchant needs. For users, 5G messaging can provide an immersive service experience, and combined with the identity authentication capabilities of the super SIM card, one-click operations from browsing to payment can be achieved.

[0044] The service platform includes an algorithm scheduling engine, which can achieve effective service recommendation predictions based on preset rules, recommendation algorithms, and large models based on the Transformer framework.

[0045] The service platform can predict matching service functions based on the location of the user's terminal and flexibly recommend various merchant services such as coupons, discount coupons, membership benefits, limited-time discounts, etc.

[0046] Figure 1c A schematic diagram of the scenario application is shown. First, the merchant can set the location range of the electronic fence in the system. Once the user terminal enters the location range of the electronic fence, the super SIM message is sent to the user terminal by combining the super SIM card with the 5G message, so that the user can perform interactive operations on the super SIM message to achieve strong user reach and one-click security services. Then, based on the user's interactive operation, the super SIM card identity verification is performed, and then the predicted service content is pushed according to the merchant's pre-configured algorithm engine to recommend it to the user through 5G messages. After the service recommendation is executed, the behavioral data of merchants and customers involved in the service recommendation process is automatically recovered and assigned to technicians. This information can be used for labeling, calibration, fine-tuning and other tasks, using the "machine + manual" method to optimize the system's basic large model and achieve model quality optimization and update.

[0047] For example, based on Figure 1cThe scenario shown. Merchants pre-configure the required information such as the electronic fence distance, super SIM card parameters, 5G message templates and algorithm models in the system. The base station data is used to sense customers who have entered the fence, and then the mobile terminal sends a request instruction to the super SIM card, and the super SIM card pops up a super SIM must-reach message on the customer's terminal according to the preset data and strategy. If the user performs an interactive operation on the super SIM message and chooses to accept the request, the identity verification is further performed through the super SIM card, and the 5G message application is called up after the verification passes, and the recommended service content is displayed through the 5G message, so that users can continue to enjoy one-click attention-free services such as browsing, recommendation, and ordering. In the process of service recommendation, the data involved will be fully collected for modeling and analysis. In addition, if the user rejects the request, a message to regain the service method can be sent, and subsequent service content recommendations will no longer be executed to avoid affecting the customer's other communication experience.

[0048] In the solution provided in the embodiment of the present application, the service content displayed through 5G messages is predicted by a pre-trained service content recommendation model. This model application method can take into account the stability and step-by-step nature of commercial use, and can provide service differentiation, and merchants can flexibly configure according to actual needs. Among them, predicting service content based on a pre-trained model and displaying it through 5G messages can be implemented based on a database and service interface, and has stability, service function differentiation and actual application economy. Compared with preset service content recommendation methods such as keyword triggering, this solution can pay attention to the preference details of users' actual applications through a pre-trained model, effectively improving the degree of matching between service content recommendations and users' own preferences.

[0049] Based on the solution provided in the above embodiment, optionally, Figure 2 As shown, before the above step S14, that is, before instructing the user terminal to display the service content matching the identity information of the super SIM card through a 5G message based on the identity verification result, it also includes: S21: Acquire service operation information of sample users, where the service operation information includes identity information of the sample users and interactive operation information of the sample users on optional service items.

[0050] In this step, the service operation information of the sample users is obtained in batches in advance. The service operation information may be obtained from a database. The service operation information includes the identity information of the sample users. The identity information may be the user's own information, the identity information of the user's super SIM card, and other information that can be used to identify the identity. The interactive operation information included in the service operation information refers to the information that the sample user performs interactive operations on the optional service items. The interactive operation information can indicate that the sample user has selected the service item required by himself from at least one optional service item, that is, the service information that the sample user has chosen to accept.

[0051] S22: constructing a training sample based on the service operation information, and generating a training label corresponding to the training sample, wherein the training label indicates whether a sample user of the corresponding training sample selects a target service item through an interactive operation.

[0052] In this step, training samples and corresponding training labels are constructed based on the above service operation information to provide sample data basis for subsequent training models. The training samples are constructed from service operation information and have the identity characteristics and interactive operation characteristics of sample users, wherein the interactive operation characteristics can specifically represent the characteristics of optional service items provided by users. The training labels corresponding to the training samples are used to indicate whether the sample users have selected the target service items. The training labels can indicate whether the identity of the user in the corresponding training sample matches the optional service items, that is, whether the optional service items are the service items that the user is willing to select.

[0053] S23: Training a transformation neural network based on the training samples and corresponding training labels to obtain a service content recommendation model.

[0054] In this step, the transformer neural network based on the Transformer framework is trained based on the above training samples and corresponding training labels, so that the model learns the correspondence between the user's identity characteristics and the interactive operations, that is, learns which service items the user's identity characteristics match, so that the trained model can predict the service items that the user is willing to accept based on the user's identity characteristics and other information.

[0055] Through the solution provided in the embodiment of the present application, the transformation neural network can be trained based on the user's identity information and interactive operation information, so that the model can learn the matching relationship between the user's identity information and the service items that the user is willing to accept, so that the service content recommendation model can accurately predict which service the user actually needs based on the user's identity information, thereby improving the effectiveness of service recommendations.

[0056] Based on the solution provided in the above embodiment, optionally, Figure 3a As shown, in the above step S23, the transformation neural network is trained based on the training samples and the corresponding training labels to obtain a service content recommendation model, including: S31: Input the service operation information into an embedding layer to obtain a service operation vector.

[0057] Figure 3b An optional model training process diagram is shown. In this step, the service operation information is input as input into the embedding layer Embedding, so that the service operation information is converted into a vector form to obtain a service operation vector.

[0058] S32: Input the service operation vector into the transformation neural network to sequentially pass through the multi-head attention layer, the first residual normalization layer, the feedforward network layer and the second residual normalization layer of the transformation neural network, and determine the model loss through the loss function to iteratively train the transformation neural network to obtain the service content recommendation model.

[0059] Then, the service operation vector is input into the transformation neural network Transfformer, and passes through the multi-head attention layer Multi-Head Attention, the first residual normalization layer ResNets 101 Norm (connected to Multi-Head Attention), the feedforward network layer Feed Forward and the second residual normalization layer ResNets 101 Norm (connected to Feed Forward) in sequence, and then the model loss is determined by the loss function. In this example, the cross entropy loss Cross EntropyLoss can be selected, so as to iteratively train the transformation neural network to obtain the service content recommendation model.

[0060] The service content recommendation model in the solution provided by the embodiment of the present application is iterated by training the model and fine-tuning the method. Since 5G messages are rich media communication messages, the core modules in this solution all use multimodal tasks and use encoders to perform sequence processing on the input data. Among them, the input data can specifically include user location, text, pictures, interactive actions and other information of the service operation information.

[0061] Among them, multimodal tasks can specifically involve M-ICL (Multimodal In-Context Learning) and M-CoT (Multimodal Chain of Thought), which are used to ensure that the model can calculate weights based on different position information when processing word positions in a sequence, thereby effectively capturing the contextual relationship between words, sentences, and paragraphs, and triggering the association mechanism of the thought chain.

[0062] Optionally, for the input data, the model can generate multiple different and reasonable responses based on contextual information, thereby simulating the nonlinear, multi-dimensional and highly flexible thinking characteristics of humans when dealing with problems.

[0063] The M-IT (Multimodal Instruction Tuning) module can be used to fine-tune the model's understanding of input generalization content and recommendation accuracy. Specifically, instruction tuning can be performed through fine-tuning methods such as reinforcement learning and reward and punishment evaluation.

[0064] Among them, the network back propagation optimization algorithm is used to update the model parameters, and the gradient change of the cross entropy loss function is selected to gradually adjust the parameters. During the training process, the gradient can be controlled to change reasonably through monitoring to avoid overfitting and avoid gradient explosion or disappearance.

[0065] In the solution provided in the embodiment of the present application, the model input layer can convert information into different vector block representations according to the rich media type, for example, Represents 5G message information. Use embedding technology to reduce and extract features, and perform z-score standardization and feature concatenation to obtain: .in, Indicates the embedding result. Represents a message vector.

[0066] Since Transformer uses global information, PE can be used to encode the position in practical applications. For example, it can be expressed as:

[0067]

[0068] Among them, d represents the dimension, and 2i and 2i+1 represent parity.

[0069] Similarly, the vector representations of the recommended services and feedback are calculated accordingly, and ResNets101 is selected as the network for extracting service information.

[0070] Through the solution provided in the embodiment of the present application, information vectorization conversion can be achieved through the embedding layer, and the service operation information characteristics can be fully extracted, which is conducive to the model learning the characteristic association relationship between the user identity and the selected service item, thereby improving the accuracy of the prediction results of the service content recommendation model.

[0071] Based on the solution provided in the above embodiment, optionally, Figure 4 As shown, before the above step S32, that is, before the service operation vector is input into the transformation neural network, it also includes: S41: Constructing a graph neural network based on the service operation information, wherein the graph neural network is used to characterize the association between the identity information of the sample user and the optional service items.

[0072] See also Figure 3bBefore the vector is input into the Transformer for encoder, it can also pass through a non-Euclidean graph network GNN (Gated Graph Neural Network). In GNN, matching nodes are built based on the user's identity information and optional service items. The user's identity information and optional service item information are stored between the nodes. There are also association elements between the identity node and the service item node, which fully reflects the association relationship between the user's identity information and the service item.

[0073] S42: Determine parameters of the multi-head attention layer based on the graph neural network.

[0074] In the embodiment of the present application, the service operation vector output by the above embedding layer is used to train the graph neural network, and then the parameters of Multi-HeadAttention are generated based on the training results of the graph neural network. Through the multi-head attention layer in the above transformation neural network, multiple independent self-attention layers are used for parallel calculation, and then the results are merged to capture information at different levels.

[0075] This parameter can be expressed as ,in, is the output of a single-layer self-attention mechanism. is the matrix concatenation transformation function, and Z is the final result.

[0076] in, The calculation formula is: ,in, is the input vector Use matrix transformation to get the result, yes The number of columns in the vector can be used to prevent the inner product from being too large.

[0077] In practical applications, the parameters of the multi-head attention layer It is sent to the residual feedforward device, and the network convergence can be accelerated by using the optimized parameter quality of GNN. Among them, Q, K, and V are obtained by linear transformation of the input vector x, which can improve the fitting ability of the model. Optionally, the above Q can refer to the information to be queried, K can refer to the queried vector, and V can refer to the value obtained by the query.

[0078] Based on the solution provided in the above embodiment, optionally, Figure 5 As shown, after the above step S32, that is, after the model loss is determined by the loss function, it also includes: S51: Iteratively training the transformation neural network based on the classification evaluation index and the model loss to obtain a service content recommendation model, wherein the classification evaluation index is determined based on a weighted harmonic mean of accuracy and recall.

[0079] See also Figure 3b In the embodiment of the present application, the F1 module can be used to evaluate content comprehension. Among them, the manual evaluation dimension and the penalty score can be combined to use a hybrid method of "machine + manual" to complete the reward and punishment of the output results. The above reward and punishment evaluation method is highly consistent with the characteristics of the task scenario, and effectively ensures the rationality and quality of the model output.

[0080] Based on the solution provided in the above embodiment, optionally, Figure 6 As shown, before the above step S51, that is, before iteratively training the transformation neural network based on the classification evaluation index and the model loss to obtain the service content recommendation model, it also includes: S61: BLEU-based evaluation index parameters 、ROUGE evaluation index parameters and preset parameters Determine the classification evaluation index for: .

[0081] In terms of information content comprehension, the solution provided in the embodiment of this application can use the BLEU indicator for evaluation. Considering the characteristics of message length, the parameter In Furthermore, considering that this indicator only focuses on accuracy, which may cause missed translations, ROUGE can also be combined to improve the recall rate.

[0082] In the embodiment of the present application, the BLEU evaluation index is combined with the ROUGE evaluation index, see Figure 3b ,when Time , and Represent the results of BLEU and ROUGE respectively.

[0083] In addition, the entire Transformer network uses a fully connected approach to perform fusion calculations, which can effectively enhance attention to message content, user portraits, service category characteristics, and other aspects, and improve the rationality of matching.

[0084] Based on the solution provided in the above embodiment, optionally, Figure 7 As shown, in the above step S23, the transformation neural network is trained based on the training samples and the corresponding training labels to obtain a service content recommendation model, including: S71: Based on the training samples and the corresponding training labels, a transformation neural network is trained through a cross entropy loss function to obtain a service content recommendation model, wherein the cross entropy loss function is:

[0085] Wherein, N represents the total number of training samples, i represents the identification number of the training sample, j represents the classification index number of the data category in the training sample, M represents the total number of data categories in the training sample, p represents the output prediction value, and y represents the training label. The training label is, for example, 0 or 1, where 1 represents relevance, that is, the user selects the corresponding service item, and 0 represents no relevance, that is, the user rejects the service item.

[0086] In the solution provided in the embodiment of the present application, since the data categories involved in the model are diverse, using the cross entropy function as an evaluation function can effectively improve the training effectiveness for multi-category information.

[0087] Based on the solution provided in the above embodiment, optionally, Figure 8a As shown, before the above step S11, that is, before obtaining the location information reported by the user terminal, it also includes: S81: Receive configuration information corresponding to an optional service item, where the configuration information includes at least one of a preset area, super SIM card parameters, and an identifier of a service content recommendation model.

[0088] See also Figure 8b In actual applications, merchants providing services can configure the relevant information of optional services in advance. Merchants can first register and log in to the service platform. During the registration process, merchants can choose to open a card management platform account and a subscription account platform account according to their own needs. For registered merchants, they can log in based on the platform account.

[0089] After successfully logging in, configure the card management platform parameters or subscription account platform parameters through the service platform. The configured parameters include, for example, the user's electronic fence range, the frequency of the Beta Super SIM message, the 5G message template, the bottom menu bar visualization settings, the ChatBot interaction settings, the algorithm scheduling strategy, etc.

[0090] The configuration results can be displayed on a visual page so that merchants can intuitively see the configuration effect. If you are not satisfied with the configuration effect, you can modify the card management platform parameters or subscription platform parameters until the configuration effect is satisfactory, then you can save the configuration and exit the service platform account.

[0091] Based on the solution provided in the above embodiment, optionally, Figure 9a As shown, in the above step S13, based on the interactive operation on the optional service item received by the user terminal, identity verification is performed through the super SIM card, including: S91: If the interactive operation indicates that the target user selects a target service item from the super SIM message, identity verification is performed on the target user through the super SIM card.

[0092] In this step, the interactive operation of the target user is parsed. If the target user selects the target service item through the operation performed on the super SIM message, the identity of the target user is further verified through the super SIM card to provide the target user with relevant service recommendations for the target service item in subsequent steps.

[0093] Combine the following Figure 9b The schematic flow chart shown further illustrates this scheme.

[0094] Merchants providing services configure parameters in advance on the card management platform and 5G message subscription platform. The service platform provides an execution basis for subsequent steps through data collection and algorithm engines.

[0095] The user terminal reports its location information based on its own location. Once the user terminal enters the merchant's electronic fence, the service platform can activate the card platform STK forced pop-up window capability and send a super SIM must-reach message to the user terminal through the machine-card channel.

[0096] After the user terminal receives and displays the Super SIM must-reach message, the user can choose to accept or reject the service item according to actual needs. If the user chooses to accept the target service item, the terminal 5G message application is called up based on the SIM authentication identity verification, and the recommended service content is displayed through the 5G message. The service content is predicted by the algorithm scheduling engine through the pre-trained service content recommendation model. In the service recommendation process, users can enjoy the services in the 5G message without registering or paying attention. Users can immerse themselves in the 5G message to experience or browse the service.

[0097] If the user chooses to reject the service, the subsequent service can be withdrawn, and a reminder to re-acquire the service will be sent to the user terminal. In actual application, if the user accidentally touches and triggers the rejection operation, the service can be obtained again through the reminder to re-acquire the service.

[0098] In actual applications, the interactive operation performed by the user can indicate whether the user has agreed to the local life information push service. If the user clicks on the Super SIM must-reach message to indicate willingness to subscribe, a receipt indicating whether the subscription is successful can be further fed back to the user terminal. If the subscription is successful, the subscription, configuration, account and other related information can be synchronized to the Super SIM platform and the 5G message subscription platform respectively to ensure that the service-related information is synchronized and consistent.

[0099] Furthermore, users can also perform operations on subscribed local life information to deeply experience local information services. For example, when users open, browse and select local life information pages, they can be guided to enter the 5G message subscription account. In the 5G message subscription account, users can deeply experience the service content by clicking the bottom menu bar, clicking the floating menu, browsing rich media messages, etc. on the display interface.

[0100] Based on the solution provided in the above embodiment, optionally, Fig.10 As shown, in the above step S14, based on the identity verification result, instructing the user terminal to display service content matching the identity information of the super SIM card through a 5G message, including: S101: If the identity verification result indicates that the identity verification of the target user is passed, and the service content matched by the target user includes transaction service content, instructing the super SIM card to provide an identity authentication function to the user terminal through a digital certificate.

[0101] See also Figure 9b In the process shown, if the service content matched by the target user includes transaction service content, such as sending red envelopes, transferring money, etc., then the authentication, payment and other functions are called based on the super SIM card, and the security transaction related capabilities of the super SIM card are called through the digital certificate. Specifically, the SIM Shield application can provide effective identity security protection for mobile payment scenarios.

[0102] In addition, in the process provided by the embodiment of the present application, if the execution step fails due to an error or other reasons, a retry may be performed based on a preset number of times, for example, three times. If the retry reaches the preset number of times and there is still an abnormality, the customer service center may be fed back and the customer may be helped to manually troubleshoot the cause of the operation failure, such as the cause of subscription failure, the cause of SIM capability call failure, etc.

[0103] In addition, when the service platform authenticates the user's identity, calls functions, obtains data, and other steps that require the user to wait, the current execution progress can be displayed to the user, and the user can be guided to wait or perform the operations required for authentication by pop-up prompts, thereby optimizing the user's visual experience.

[0104] The solution provided in the embodiment of the present application combines the technical advantages of super SIM cards, 5G messages, and large models. It uses super SIM cards and 5G messages to achieve strong user reach and service interaction, and uses large model iterations to continuously evolve service recommendation capabilities.

[0105] Based on the super SIM card's must-reach message function, it can effectively receive the user's feedback receipt for the super SIM message, and instantly know whether the user accepts or rejects the service item. For the target service items selected by the user, the super SIM is used to communicate with the 5G message gateway, and the static data of the big network is combined to realize the unique mapping between terminal information, mobile phone number, service system and identification, which can effectively realize the user's login identity verification. In transaction scenarios involving user sensitive identity information, the SIM Shield service function of the super SIM card is used to apply digital certificates to realize reliable verification of user identity, and multiple information verifications such as the three elements of the operator, the transaction party interface, and the message service identification are combined to realize the security of the payment scenario.

[0106] Based on the big model function, the service content recommendation model is trained through the Transformer framework, and multimodal M-ICL, M-CoT and M-IT modules are expanded to fine-tune the model parameters by imitating human reinforcement learning and reward and punishment strategies. Among them, the output content can be verified by manual or professional technical data to realize the industry big model application of "machine + manual" integration.

[0107] For merchants providing services, the solution provided by the embodiment of the present application uses super SIM messages and 5G messages to implement service recommendations, which can ensure that messages are delivered and avoid being drowned in a large amount of invalid message content. And through super SIM messages, user feedback operations can be effectively obtained, which is conducive to merchants adjusting service provision strategies according to the actual operation results of users. In addition, the pre-trained service content recommendation model can effectively predict the service content required by users, which can reduce the cost of merchants recommending services and improve service conversion rates.

[0108] This solution enables users to intuitively and effectively see the service content through super SIM messages and 5G messages. Users do not need to follow the official account or log in with identity information. This allows users to experience services quickly and conveniently, reducing the complexity of service experience operations.

[0109] In addition, this solution uses the location information reported by the user terminal and authorized by the user himself, and can perform service recommendations based on the actual effective location information of the user terminal.

[0110] In order to solve the problems existing in the related art, the embodiment of the present application further provides a service recommendation device 110, such as Fig.11 As shown, including: The acquisition module 111 acquires the location information reported by the user terminal, where the user terminal is pre-installed with a super SIM card; The instructing module 112 instructs the user terminal to display a super SIM message if the location information indicates that the user terminal is located within a preset electronic fence, wherein the super SIM message includes information about optional service items corresponding to the preset electronic fence; A verification module 113, which performs identity verification through the super SIM card based on the interactive operation on the optional service item received by the user terminal; The display module 114 instructs the user terminal to display service content matching the identity information of the super SIM card through a 5G message based on the identity verification result, wherein the service content is the service content predicted by the pre-trained service content recommendation model based on the identity information.

[0111] Through the device provided by the embodiment of the present application, first, the location information reported by the user terminal is obtained, and the user terminal is pre-installed with a super SIM card; if the location information indicates that the user terminal is located within a preset electronic fence, the user terminal is instructed to display a super SIM message, wherein the super SIM message includes information about optional service items corresponding to the preset electronic fence; then, based on the interactive operation of the optional service items received by the user terminal, identity verification is performed through the super SIM card; finally, based on the identity verification result, the user terminal is instructed to display service content matching the identity information of the super SIM card through a 5G message, wherein the service content is the service content predicted by the pre-trained service content recommendation model based on the identity information. Through this solution, the information of the optional service items is accurately provided to the user based on the location information reported by the user terminal, and then a reliable and authentic identity verification is performed through the super SIM card. Subsequently, the display is performed through the 5G message, which can improve the richness of the message content, realize multi-form display of the service content matching the identity information, and optimize the readability of the service content. Among them, the service content is predicted by the pre-trained service content recommendation model based on the identity information, which can improve the matching degree between the displayed service content and the identity information of the super SIM card, and enhance the attractiveness of the recommended service to the user.

[0112] Among them, the above modules in the device provided by the embodiment of the present application can also implement the method steps provided by the above method embodiment. Alternatively, the device provided by the embodiment of the present application can also include other modules in addition to the above modules to implement the method steps provided by the above method embodiment. And the device provided by the embodiment of the present application can achieve the technical effects that can be achieved by the above method embodiment.

[0113] Preferably, an embodiment of the present application also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, each process of the above-mentioned service recommendation method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0114] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above-mentioned service recommendation method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0115] An embodiment of the present application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. The computer program can be operated to enable a computer to execute part or all of the steps of the above-mentioned service recommendation method embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0116] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0117] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate a device for implementing the functions specified in one process or multiple processes in the flowchart and / or one box or multiple boxes in the block diagram.

[0118] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0119] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0120] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0121] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0122] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0123] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0124] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0125] The above is only the embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A service recommendation method, characterized in that: include: Acquiring location information reported by a user terminal, wherein the user terminal is pre-installed with a super SIM card; If the location information indicates that the user terminal is located within a preset electronic fence, instructing the user terminal to display a super SIM message, where the super SIM message includes information about optional service items corresponding to the preset electronic fence; Based on the interactive operation on the optional service item received by the user terminal, performing identity verification through the super SIM card; Based on the identity verification result, the user terminal is instructed to display service content matching the identity information of the super SIM card through a 5G message, wherein the service content is the service content predicted by a pre-trained service content recommendation model based on the identity information.

2. The method according to claim 1, characterized in that Before instructing the user terminal to display service content matching the identity information of the super SIM card through a 5G message based on the identity verification result, the method further includes: Acquire service operation information of the sample user, wherein the service operation information includes identity information of the sample user and interactive operation information of the sample user on optional service items; Building a training sample based on the service operation information, and generating a training label corresponding to the training sample, wherein the training label indicates whether a sample user of the corresponding training sample selects a target service item through an interactive operation; The transformation neural network is trained based on the training samples and the corresponding training labels to obtain a service content recommendation model.

3. The method according to claim 2, characterized in that Training a transformation neural network based on the training samples and corresponding training labels to obtain a service content recommendation model includes: Inputting the service operation information into an embedding layer to obtain a service operation vector; The service operation vector is input into the transformation neural network to sequentially pass through the multi-head attention layer, the first residual normalization layer, the feedforward network layer and the second residual normalization layer of the transformation neural network, and the model loss is determined by the loss function to iteratively train the transformation neural network to obtain the service content recommendation model.

4. The method according to claim 3, characterized in that Before inputting the service operation vector into the transformation neural network, the method further includes: Constructing a graph neural network based on the service operation information, wherein the graph neural network is used to characterize the association between the identity information of the sample user and the optional service items; The graph neural network is used to determine parameters of the multi-head attention layer.

5. The method according to claim 3, characterized in that After determining the model loss through the loss function, it also includes: The transformation neural network is iteratively trained based on a classification evaluation index and the model loss to obtain a service content recommendation model, wherein the classification evaluation index is determined based on a weighted harmonic mean of precision and recall.

6. The method according to claim 5, characterized in that Before iteratively training the transformation neural network based on the classification evaluation index and the model loss to obtain the service content recommendation model, the method further includes: Based on BLEU evaluation index parameters 、ROUGE evaluation index parameters and preset parameters Determine the classification evaluation index for: .

7. The method according to claim 5, characterized in that Training a transformation neural network based on the training samples and corresponding training labels to obtain a service content recommendation model includes: Based on the training samples and the corresponding training labels, the transformation neural network is trained by a cross entropy loss function to obtain a service content recommendation model, wherein the cross entropy loss function is: Among them, N represents the total number of training samples, i represents the identification number of the training sample, j represents the classification index number of the data category in the training sample, M represents the total number of data categories in the training sample, p represents the output prediction value, and y represents the training label.

8. The method according to any one of claims 1 to 7, characterized in that Before obtaining the location information reported by the user terminal, the following steps are also included: Configuration information corresponding to the optional service item is received, where the configuration information includes at least one of a preset area, a super SIM card parameter, and an identifier of a service content recommendation model.

9. The method according to any one of claims 1 to 7, characterized in that: Based on the interactive operation on the optional service item received by the user terminal, performing identity verification through the super SIM card includes: If the interactive operation represents that the target user selects a target service item from the super SIM message, identity verification is performed on the target user through the super SIM card.

10. The method according to any one of claims 1 to 7, characterized in that: Instructing the user terminal to display service content matching the identity information of the super SIM card through a 5G message based on the identity verification result, including: If the identity verification result indicates that the identity verification of the target user is passed, and the service content matched by the target user includes transaction service content, the super SIM card is instructed to provide an identity authentication function to the user terminal through a digital certificate.

Images