Host card emulation service routing method and device, storage medium and computer device

Abstract

The host card simulation service routing method, apparatus, storage medium, and computer equipment provided in this application, when detecting a target user performing an HCE flashing operation, first collect the target user's usage environment data and determine the service routing model corresponding to the application identifier used by the target user. This model adopts a dual-path fusion prediction mechanism of similarity retrieval and usage context inference, which can simultaneously combine scene feature matching and user usage habits to infer accurate target service IDs from usage environment data, thereby improving the accuracy of routing judgment and avoiding routing conflicts of multiple services under the same AID. Subsequently, the HCE service corresponding to the target service ID is called to respond to the flashing operation, ensuring the accuracy of service routing, and user profile samples are generated based on the flashing results for weight optimization of the service routing model, enabling the model to continuously learn the user's personalized usage preferences and scene change patterns, and achieve dynamic self-evolution of the model.

Description

Technical Field

[0001] This application relates to the field of data communication technology, and in particular to a host card emulation service routing method, apparatus, storage medium and computer equipment. Background Technology

[0002] With the rapid development of mobile payment and NFC near-field communication technologies, Host Card Emulation (HCE) technology has been widely used in various mobile terminals due to its advantage of enabling card swiping without relying on secure elements. In practical applications, a mobile terminal can integrate multiple HCE services, such as simultaneously supporting multiple functions like public transport cards, bank cards, and access control cards. When the terminal triggers an HCE swipe, it quickly routes to the corresponding HCE service through an application identifier.

[0003] However, in practical applications, to reduce the complexity of application development, developers often bind multiple HCE services under the same application identifier. While this simplifies the development process, it introduces a serious problem: when a terminal initiates a card simulation request, because multiple HCE services share the same AID, the system cannot accurately determine which HCE service to route to, leading to routing conflicts and failures, requiring the user to manually reselect the service. Summary of the Invention

[0004] The purpose of this application is to at least solve one of the aforementioned technical defects, particularly the technical defect in the prior art where multiple HCE services share the same AID, and the system cannot accurately determine which HCE service to route to, thus leading to routing conflicts and failures.

[0005] This application provides a method for simulating service routing on a host card, the method comprising:

[0006] When a target user is detected performing an HCE flashing operation, the user's usage environment data is collected.

[0007] Determine the service routing model corresponding to the application identifier used by the target user, and use the service routing model to perform dual-path fusion prediction of similarity retrieval and usage context reasoning on the usage environment data to obtain the target service ID;

[0008] The HCE service corresponding to the target service ID is invoked to respond to the flashing operation, the flashing result is obtained, and a user profile sample is generated based on the flashing result;

[0009] The service routing model is weighted and optimized using the user profile samples to obtain the latest service routing model for the application identifier.

[0010] Optionally, the collection of the target user's usage environment data includes:

[0011] Collect the target user's current geographical location and current timestamp at the time of flashing;

[0012] The target user's usage environment data is generated based on the current geographical location and the current timestamp.

[0013] Optionally, the process of constructing the service routing model includes:

[0014] Obtain all HCE services under the application identifier and generate scenario description text for each HCE service;

[0015] A vector database of application identifiers is constructed based on the description text of each scenario, and the usage environment data of the target user for each HCE service in a historical time period is collected to form a personal profile database.

[0016] The preset large language model is fine-tuned and trained using the vector database and the personal profile database to obtain the service routing model corresponding to the application identifier.

[0017] Optionally, the step of constructing the vector database of application identifiers based on the description text of each scenario includes:

[0018] The text describing each scene is segmented into words to obtain the segmentation results, and the core feature words of each scene description text are identified from the segmentation results.

[0019] Each core feature word is converted into a scenario description vector of a preset dimension, and each feature embedding vector is bound and stored with the service ID of the corresponding HCE service to form a vector database corresponding to the application identifier.

[0020] Optionally, the service routing model includes a similarity retrieval network, a contextual reasoning network, and a fusion decision network;

[0021] The step of using the service routing model to perform dual-path fusion prediction on the usage environment data through similarity retrieval and usage context reasoning to obtain the target service ID includes:

[0022] The similarity retrieval network is used to generate the current environment vector of the used environment data, and the current environment vector is matched with each scene description vector in the vector database to obtain the scene matching result;

[0023] The usage environment data is input into the contextual reasoning network, so that the contextual reasoning network performs intent reasoning on the usage environment data based on the personal profile database and outputs the intent reasoning result;

[0024] The target service ID is obtained by fusing the scene matching result and the intent reasoning result through the fusion decision network.

[0025] Optionally, the step of fusing the scene matching result and the intent reasoning result through the fusion decision network to obtain the target service ID includes:

[0026] The fusion decision network is used to determine whether the prediction service ID in the scene matching result and the intent reasoning result are consistent.

[0027] If so, the predicted service ID is output through the fusion decision network as the target service ID;

[0028] If not, when the confidence level of the intent reasoning result exceeds a preset threshold, the predicted service ID of the intent reasoning result is output through the fusion decision network as the target service ID.

[0029] When the predicted service ID in the scene matching result and the intent reasoning result are inconsistent and the confidence level of the intent reasoning result does not exceed the preset threshold, a pop-up confirmation mode is triggered so that the intent reasoning result outputs the target service ID based on the confirmation result of the target user.

[0030] Optionally, generating a user profile sample based on the flashing result includes:

[0031] When the flashing result is successful, a user profile sample is generated based on the usage environment data and the target service ID;

[0032] When the flashing process fails, the system monitors the correction service ID selected by the target user and generates a user profile sample based on the usage environment data, the target service ID, and the correction service ID.

[0033] This application also discloses a host card emulation service routing device, comprising:

[0034] The data acquisition module is used to collect the target user's usage environment data when the target user performs an HCE flashing operation.

[0035] The model prediction module is used to determine the service routing model corresponding to the application identifier used by the target user, and to use the service routing model to perform dual-path fusion prediction of similarity retrieval and usage context reasoning on the usage environment data to obtain the target service ID.

[0036] The sample generation module is used to call the HCE service corresponding to the target service ID to respond to the flashing operation, obtain the flashing result, and generate a user profile sample based on the flashing result;

[0037] The model optimization module is used to optimize the service routing model with weights using the user profile samples to obtain the latest service routing model for the application identifier.

[0038] This application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the host card emulation service routing method as described in any of the above embodiments.

[0039] This application also discloses a computer device comprising: one or more processors, and a memory;

[0040] The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the host card emulation service routing method as described in any of the above embodiments.

[0041] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0042] The host card simulation service routing method, apparatus, storage medium, and computer equipment provided in this application, when detecting a target user performing an HCE flashing operation, can first collect the target user's usage environment data to enrich the decision basis for service routing from the data source. Then, it can determine the service routing model corresponding to the application identifier used by the target user. This model adopts a dual-path fusion prediction mechanism of similarity retrieval and usage context inference, which can simultaneously combine scene feature matching and user usage habits to infer accurate target service IDs from usage environment data. Compared with single matching logic, it significantly improves the accuracy of routing judgment and avoids routing conflicts of multiple services under the same AID. Subsequently, it can call the HCE service corresponding to the target service ID to respond to the flashing operation, ensuring the accuracy of service routing. Then, it can generate user profile samples based on the flashing results, providing real and reliable training materials for continuous model optimization. That is, it uses the user profile samples to optimize the weights of the service routing model, enabling the model to continuously learn the user's personalized usage preferences and scene change patterns, and realize the dynamic self-evolution of the model. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating a host card emulation service routing method provided in an embodiment of this application;

[0045] Figure 2 A flowchart illustrating the process of constructing a service routing model as provided in this application embodiment;

[0046] Figure 3 This is a schematic diagram of the structure of a host card emulation service routing device provided in an embodiment of this application;

[0047] Figure 4 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0048] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0049] In practical applications, to reduce the complexity of application development, developers often bind multiple HCE services under the same application identifier. While this simplifies the development process, it introduces a serious problem: when a terminal initiates a card simulation request, because multiple HCE services share the same AID, the system cannot accurately determine which HCE service to route to, leading to routing conflicts and failures, requiring the user to manually reselect the service.

[0050] Based on this, this application proposes the following technical solution, as detailed below:

[0051] In one embodiment, such as Figure 1 As shown, Figure 1 This is a flowchart illustrating a host card emulation service routing method provided in an embodiment of this application. The present application provides a host card emulation service routing method, specifically including the following:

[0052] S110: When the target user is detected performing HCE flashing, collect the target user's usage environment data.

[0053] In this step, when the computer device detects that the target user is performing an HCE flashing operation, it can first collect the target user's usage environment data, such as time, location, network environment, etc., to enrich the decision basis for service routing from the data source.

[0054] Specifically, usage environment data refers to information related to the objective environment in which a user performs an HCE flashing operation. This information reflects the user's current card-swiping scenario and is an important basis for subsequent service routing decisions. By using environment data, computer devices can describe the target user's usage habits during the current flashing operation, providing reliable data input for subsequent predictive inference.

[0055] S120: Determine the service routing model corresponding to the application identifier used by the target user, and use the service routing model to perform dual-path fusion prediction of similarity retrieval and usage context reasoning on the usage environment data to obtain the target service ID.

[0056] In this step, after collecting the target user's usage environment data through step S110, the computer device can determine the service routing model corresponding to the application identifier used by the target user. This model adopts a dual-path fusion prediction mechanism of similarity retrieval and usage context reasoning, which can simultaneously combine scene feature matching and user usage habits to reason about the usage environment data and obtain an accurate target service ID. Compared with a single matching logic, it can significantly improve the accuracy of routing judgment and avoid routing conflicts of multiple services under the same AID.

[0057] Among them, the service routing model refers to the prediction model pre-trained for multiple HCE services under the application identifier. It can infer the target HCE service that best meets the user's needs in the current scenario based on the user's current usage environment data, providing model support for quickly and accurately completing service routing.

[0058] Specifically, the service routing model can employ a dual-path fusion prediction mechanism that combines similarity retrieval and usage context reasoning. The similarity retrieval path is mainly used to perform feature matching analysis between the current usage environment data and pre-stored service scenario features, while the usage context reasoning path is mainly used to infer and judge each HCE service by combining the target user's historical usage habits, scenario preferences, and behavioral patterns, thereby achieving joint prediction of the target service ID.

[0059] Furthermore, the service routing model can also perform fusion calculations on the scene matching results output by the similarity retrieval path and the behavioral preference results output by the contextual reasoning path, and determine the HCE service with the highest confidence based on the calculation results, and output the service ID corresponding to the HCE service as the target service ID.

[0060] Understandably, compared with the traditional method of matching a single AID or judging by fixed rules, the method of this application, which combines scene feature matching ability and user behavior reasoning ability to jointly analyze the current usage environment data, can more accurately distinguish multiple HCE services corresponding to the same AID, effectively reduce service routing conflicts and mismatch problems, improve the accuracy of automatic identification of HCE services and the success rate of flashing, and reduce the user's burden of manually switching services.

[0061] S130: Call the HCE service corresponding to the target service ID to respond to the flashing operation, obtain the flashing result, and generate a user profile sample based on the flashing result.

[0062] In this step, after determining the target service ID through step S120, the computer device can call the HCE service corresponding to the target service ID to respond to the flashing operation, ensuring the accuracy of service routing. Then, user profile samples can be generated based on the flashing results, providing real and reliable training materials for the continuous optimization of the model.

[0063] Among them, the user profile sample can represent the HCE service usage records of the target user under specific time, location and other scenario conditions. It is stored in the target user's personal profile database. Through this personal profile database, this application can analyze the target user's frequently used service types, service switching patterns in different scenarios, card swiping preferences in different time periods and the probability distribution of various HCE services, etc., to provide a data foundation for the subsequent scenario reasoning and continuous optimization of the service routing model.

[0064] Specifically, when calling the HCE service, the computer device can establish a corresponding service call link based on the target service ID and send a card simulation request to the target HCE service, enabling the target HCE service to complete the corresponding card data interaction, identity authentication, and card swiping business processing based on the NFC communication protocol.

[0065] For example, when the target service ID corresponds to the public transport card HCE service, the computer device can call the card simulation module corresponding to the public transport card to perform near-field communication interaction with the gate device; when the target service ID corresponds to the bank card payment service, the computer device can call the corresponding payment HCE service to complete payment authentication and transaction processing; when the target service ID corresponds to the access control card service, the computer device can call the access control HCE service corresponding to complete identity verification and access control opening operations.

[0066] Furthermore, after completing the HCE service call, the computer device can also collect and record the interaction results during the flashing process to generate user profile samples. The flashing results can include information such as whether the card swipe was successful, whether the user manually switched, the type of HCE service ultimately invoked, and the corresponding scenario characteristic information.

[0067] S140: Use user profile samples to optimize the service routing model with weights, and obtain the service routing model with the latest application identifier.

[0068] In this step, after obtaining the user profile sample of the current flashing operation through step S130, the computer device can use the user profile sample to optimize the weight of the service routing model, so that the model can continuously learn the user's personalized usage preferences and the changing patterns of the scenario, and realize the dynamic self-evolution of the model.

[0069] Specifically, computer devices can input user profile samples into the service routing model for incremental training to update the feature weight parameters in the service routing model, enabling the model to gradually establish the correlation between user behavior features and target HCE services. In more detail, the service routing model can use user profile samples to detect whether the HCE service ultimately triggered by the target user matches the target service ID inferred by the service routing model, and optimize the weights based on the detection results.

[0070] For example, when service consistency is detected, the service routing model can positively enhance the feature weights of the inference path corresponding to the current usage scenario. For instance, it can increase the association weights of current time features, location features, terminal state features, and user behavior features in the prediction of the corresponding HCE service, and increase the matching confidence between the current scenario and the target HCE service. If inconsistency is detected, the computer device can downgrade the feature weights of the inference path corresponding to the current usage scenario, while simultaneously establishing or enhancing the association between the current usage scenario and the final HCE service.

[0071] Understandably, this application can form a continuously iterative human-machine loop learning mechanism through "automatic reasoning—sample feedback—weight correction—retraining". In the process of multiple rounds of human-machine loop learning, the service routing model can continuously correct erroneous reasoning paths and continuously strengthen the mapping relationship between correct scenarios and corresponding HCE services, and finally gradually determine the most suitable HCE service for different use cases.

[0072] In the above embodiments, when a target user is detected performing an HCE flashing operation, the user's usage environment data can be collected first to enrich the decision basis for service routing from the data source. Then, the service routing model corresponding to the application identifier used by the target user can be determined. This model adopts a dual-path fusion prediction mechanism of similarity retrieval and usage context reasoning, which can simultaneously combine scene feature matching and user usage habits to reason about the usage environment data to obtain an accurate target service ID. Compared with a single matching logic, this significantly improves the accuracy of routing judgment and avoids routing conflicts of multiple services under the same AID. Subsequently, the HCE service corresponding to the target service ID can be called to respond to the flashing operation to ensure the accuracy of service routing. Then, user profile samples can be generated based on the flashing results to provide real and reliable training materials for the continuous optimization of the model. That is, the service routing model is optimized by using the user profile samples, so that the model can continuously learn the user's personalized usage preferences and scene change patterns, and realize the dynamic self-evolution of the model.

[0073] In one embodiment, the process of collecting the target user's usage environment data in step S110 may include:

[0074] S111: Collect the target user's current geographical location and current timestamp at the moment of flashing.

[0075] S112: Generate the target user's usage environment data based on the current geographical location and current timestamp.

[0076] In this embodiment, when collecting usage environment data, the computer device can first collect the target user's current geographical location and current timestamp at the time of flashing the device, and then generate the target user's usage environment data based on the current geographical location and current timestamp.

[0077] Specifically, the computer device can obtain the target user's current geographical location through the positioning module in the user's terminal. This geographical location can directly reflect the type of scene the user is currently in, such as a bus stop, subway station, office building area, residential area, shopping mall, school, or parking lot. At the same time, the computer device can also obtain the current timestamp corresponding to the current flashing time through the system clock module, including the current date, current time, weekday information, holiday information, and time period type. This timestamp can also be divided into scene time tags such as weekday morning peak hours, lunchtime consumption hours, evening commuting hours, or nighttime access control hours according to actual needs.

[0078] Furthermore, the computer device can structurally encapsulate the collected current geographical location and current timestamp to generate usage environment data. Of course, the usage environment data in this application is not limited to geographical location and timestamp, but may also include network environment, terminal connection status, and user activity status, etc., without limitation.

[0079] Understandably, by jointly analyzing various parameters in environmental data, computer devices can generate more granular scene semantic information; for example, a target user might be swiping their card in an office building area during weekday morning rush hour. Based on this, since the same geographical location may correspond to different HCE service needs at different times, computer devices can improve scene recognition accuracy through joint reasoning based on time and location. For instance, an office building area might correspond to access control card services during weekday morning rush hour, while during lunchtime it might correspond to payment-related HCE services.

[0080] In one embodiment, such as Figure 2 As shown, Figure 2 A flowchart illustrating the process of constructing a service routing model as provided in this application embodiment; Figure 2 In step S120, the process of constructing the service routing model may include:

[0081] S121: Obtain all HCE services under the application identifier and generate scenario description text for each HCE service.

[0082] S122: Construct a vector database of application identifiers based on the description text of each scenario, and collect the usage environment data of the target user for each HCE service in a historical time period to form a personal profile database.

[0083] S123: Use vector databases and personal profile databases to fine-tune and train the preset large language model to obtain the service routing model corresponding to the application identifier.

[0084] In this embodiment, when constructing the service routing model, the computer device can first obtain all HCE services under the application identifier and generate scenario description text for each HCE service. Then, it can construct a vector database of application identifiers based on each scenario description text. Additionally, it can collect the usage environment data of target users for each HCE service in historical time periods to form a personal profile database. Then, it can use the vector database and the personal profile database to fine-tune and train the preset large language model to obtain the service routing model corresponding to the application identifier.

[0085] Specifically, for all HCE services of different types simultaneously bound to an application identifier, the computer device can generate scenario description text for each HCE service. This scenario description text can describe the typical use cases, triggering environments, behavioral characteristics, and business attribute information of the corresponding HCE service. For example, the scenario description text for a bus card HCE service could be "used at bus stops during weekday morning and evening rush hours for bus fare payment," while the scenario description text for an access card HCE service could be "used in office building areas during weekday morning and evening commuting hours for access control and identity verification." Subsequently, the computer device can vectorize each scenario description text to generate a scenario description vector for each HCE service, thereby constructing a vector database to provide a matching basis for subsequent similarity retrieval paths.

[0086] At the same time, computer devices can also extract the target user's usage environment data for each HCE service from the historical flashing records in the recent period, and organize and store this data to form a personal profile library of the target user. This allows the model to learn the user's unique usage habits and preferences, and provide personalized user data for subsequent usage context inference paths.

[0087] Furthermore, computer devices can use the scene matching capabilities provided by the vector database as a basis, combined with historical usage data in the personal profile database, to fine-tune and train a pre-set large language model. This allows the large language model to simultaneously master scene feature matching logic and user personalized usage reasoning logic, ultimately obtaining a service routing model that adapts to the application identifier and the corresponding user.

[0088] In one embodiment, the process of constructing a vector database of application identifiers based on the description text of each scenario in step S122 may include:

[0089] S1221: Perform word segmentation on the description text of each scene to obtain the word segmentation results, and identify the core feature words of each scene description text from the word segmentation results.

[0090] S1222: Convert each core feature word into a scenario description vector of a preset dimension, and bind and store each feature embedding vector with the service ID of the corresponding HCE service to form a vector database corresponding to the application identifier.

[0091] In this embodiment, when constructing the vector database, the computer device can first perform word segmentation on the description text of each scene to obtain the word segmentation results. Then, it can identify the core feature words of each description text from the word segmentation results and convert each core feature word into a scene description vector of a preset dimension. In this way, each feature embedding vector can be bound and stored with the service ID of the corresponding HCE service to form a vector database corresponding to the application identifier, providing a matching basis for subsequent similarity retrieval paths.

[0092] Specifically, during word segmentation, the computer device can segment the scene description text into phrases based on part-of-speech, word frequency, and contextual semantics using word segmentation tools. It then selects core feature words related to scene characteristics and filters out meaningless conjunctions, auxiliary words, and other redundant components. Subsequently, the computer device can convert the extracted core feature words into scene description vectors of a unified preset dimension using a pre-defined word vector conversion model, completing feature embedding. Finally, each obtained scene description vector is bound to the service ID of the corresponding HCE service and stored uniformly in a vector database of the corresponding application identifier, completing the construction of the vector database.

[0093] In one embodiment, the service routing model in step S120 may include a similarity retrieval network, a contextual reasoning network, and a fusion decision network; wherein, the process of using the service routing model to perform dual-path fusion prediction of similarity retrieval and contextual reasoning on usage environment data to obtain the target service ID may include:

[0094] S124: Generate the current environment vector using the environment data by using the similarity retrieval network, and perform similarity matching between the current environment vector and each scene description vector in the vector database to obtain the scene matching result.

[0095] S125: Input the usage environment data into the contextual reasoning network so that the contextual reasoning network can perform intent reasoning based on the personal profile database and output the intent reasoning result.

[0096] S126: The target service ID is obtained by fusing the scene matching results and intent reasoning results through a fusion decision network.

[0097] In this embodiment, the service routing model may include a similarity retrieval network, a contextual reasoning network, and a fusion decision network. After the computer device inputs usage environment data into the service routing model, the similarity retrieval network can generate a current environment vector of the usage environment data and perform similarity matching between the current environment vector and each scene description vector in the vector database to obtain a scene matching result; simultaneously, the contextual reasoning network can perform intent reasoning on the usage environment data based on a personal profile database and output an intent reasoning result; finally, the fusion decision network can perform a fusion decision on the scene matching result and the intent reasoning result to obtain the target service ID.

[0098] Specifically, after receiving usage environment data, the similarity retrieval network can first perform feature encoding on the usage environment data to generate a corresponding current environment vector. Then, it can use cosine similarity calculation, Euclidean distance calculation, or nearest neighbor vector retrieval algorithms to calculate the semantic distance between the current environment vector and each scene description vector, and output the corresponding scene matching result. For example, when the similarity between the current environment vector and the scene description vector corresponding to "weekday + subway station + commuting scenario" is the highest, the similarity retrieval network can output a scene matching result corresponding to the public transport card HCE service; when the current environment vector is closer to "office area + working hours + access control verification scenario", it can output a scene matching result corresponding to the access control card HCE service.

[0099] Meanwhile, the contextual reasoning network can retrieve historical usage environment data for each HCE service corresponding to the user from the personal profile database. It then combines this data with current usage environment data to statistically analyze the target user's historical confidence levels when calling each HCE service in the current usage scenario, and outputs the HCE service with the highest confidence level as the intent reasoning result. For example, when the system detects that the target user has the highest confidence level when using the public transport card service on a weekday morning, the contextual reasoning network can output the intent reasoning result corresponding to the public transport card service; similarly, when the target user has the highest confidence level when using the access control card service in the office area, it can output the intent reasoning result corresponding to the access control card service.

[0100] Finally, the fusion decision network can perform joint analysis based on the scenario matching results and intent reasoning results, and select the service ID of the corresponding HCE service as the final predicted target service ID output based on the analysis results.

[0101] In one embodiment, the process of fusing the scene matching result and the intent reasoning result through a fusion decision network to obtain the target service ID in step S126 may include:

[0102] S1261: Use a fusion decision network to determine whether the predicted service ID in the scene matching result and the intent reasoning result are consistent.

[0103] S1262: If so, the predicted service ID is output through the fusion decision network as the target service ID.

[0104] S1263: If not, when the confidence level of the intent reasoning result exceeds the preset threshold, the predicted service ID of the intent reasoning result is output through the fusion decision network as the target service ID.

[0105] S1264: When the predicted service ID in the scene matching result and the intent reasoning result are inconsistent and the confidence level of the intent reasoning result does not exceed the preset threshold, a pop-up confirmation mode is triggered so that the intent reasoning result outputs the target service ID based on the confirmation result of the target user.

[0106] In this embodiment, during the fusion decision-making process, the service routing model can use the fusion decision network to determine whether the predicted service ID in the scene matching result and the intent reasoning result are consistent. If they are, the predicted service ID can be directly output as the target service ID. Otherwise, if the confidence level of the intent reasoning result exceeds a preset threshold, the predicted service ID of the intent reasoning result is output through the fusion decision network as the target service ID. Furthermore, if the predicted service ID in the scene matching result and the intent reasoning result are inconsistent and the confidence level of the intent reasoning result does not exceed a preset threshold, a pop-up confirmation mode can be triggered so that the intent reasoning result outputs the target service ID based on the target user's confirmation result.

[0107] Specifically, when the predicted service IDs given by the scene matching result and the intent inference result are consistent, it indicates that the judgment results of these two paths can mutually verify each other, and the prediction reliability is high. Therefore, the model can directly output the consistent service ID as the target service ID. When the two results are inconsistent, it indicates that there may be multiple HCE services competing for the same AID in the current usage scenario, or the current scenario belongs to a complex and ambiguous scenario. In this case, the fusion decision network can prioritize the user's historical usage habits for judgment, that is, judge the confidence level corresponding to the intent inference result.

[0108] Specifically, when the confidence level of the HCE service in the intent reasoning result is higher than the preset threshold, it indicates that the model has a high grasp of the user's current usage intent. Therefore, the predicted service ID corresponding to the intent reasoning result is output based on the user's personalized usage habits. When the confidence level does not exceed the preset threshold, it indicates that the model cannot give a high-confidence judgment result based on the existing data. At this time, a pop-up confirmation interface can be triggered on the terminal to display multiple candidate HCE services to the target user, who can manually select the target service to be called. This dynamic fusion mechanism that combines automatic decision-making with manual confirmation can enhance the system's adaptability to complex dynamic scenarios and changes in user behavior. This not only avoids incorrect routing matching but also ensures that the user's flashing operation can be completed smoothly.

[0109] In one embodiment, the process of generating a user profile sample based on the flashing result in step S130 may include:

[0110] S131: When the flashing result is successful, generate a user profile sample based on the usage environment data and the target service ID.

[0111] S132: When the flashing result fails, monitor the correction service ID selected by the target user and generate a user profile sample based on the usage environment data, the target service ID, and the correction service ID.

[0112] In this embodiment, after the computer device calls the HCE service to perform a flashing operation, if the flashing result is successful, the computer device can generate a user profile sample based on the usage environment data and the target service ID; if the flashing result fails, the computer device can monitor the correction service ID selected by the target user and generate a user profile sample based on the usage environment data, the target service ID, and the correction service ID.

[0113] Specifically, when the flashing result is successful, it means that the target service ID output by the service routing model is consistent with the actual usage needs of the target user. That is, the current automatic routing result can correctly match the actual flashing scenario of the target user. Therefore, the computer device can directly generate the corresponding user profile sample based on the current usage environment data and the target service ID to improve the prediction weight of the path corresponding to the current usage scenario.

[0114] When the flashing process fails, it indicates that the target service ID output by the current service routing model does not match the actual flashing scenario of the target user. In this case, the computer device can trigger a service correction process. For example, it can display a list of candidate HCE services for the target user to reselect, then monitor the target user's final selected correction service ID in real time, and call the HCE service corresponding to that correction service ID to re-execute the card-swiping operation. When the target user's reselected correction service ID successfully completes the card swipe, the computer device can determine that the current correction service ID better matches the target user's actual usage intent in the current environment. At this point, the computer device can generate a corresponding user profile sample based on the usage environment data of the current scenario, the target output by the service routing model, and the correction service ID manually switched by the user. This sample is used to dynamically adjust the prediction weights of the service routing model for the corresponding path in the current usage scenario.

[0115] The host card emulation service routing apparatus provided in the embodiments of this application is described below. The host card emulation service routing apparatus described below and the host card emulation service routing method described above can be referred to in correspondence.

[0116] In one embodiment, such as Figure 3 As shown, Figure 3 This is a schematic diagram of a host card emulation service routing device provided in an embodiment of this application. This application also provides a host card emulation service routing device, including a data acquisition module 210, a model prediction module 220, a sample generation module 230, and a model optimization module 240, specifically comprising the following:

[0117] The data acquisition module 210 is used to collect the target user's usage environment data when the target user is detected performing an HCE flashing operation.

[0118] The model prediction module 220 is used to determine the service routing model corresponding to the application identifier used by the target user, and to use the service routing model to perform dual-path fusion prediction of similarity retrieval and usage context reasoning on usage environment data to obtain the target service ID.

[0119] The sample generation module 230 is used to call the HCE service corresponding to the target service ID to respond to the flashing operation, obtain the flashing result, and generate user profile samples based on the flashing result.

[0120] The model optimization module 240 is used to optimize the service routing model with weights using user profile samples to obtain the latest service routing model for application identifiers.

[0121] In the above embodiments, when a target user is detected performing an HCE flashing operation, the user's usage environment data can be collected first to enrich the decision basis for service routing from the data source. Then, the service routing model corresponding to the application identifier used by the target user can be determined. This model adopts a dual-path fusion prediction mechanism of similarity retrieval and usage context reasoning, which can simultaneously combine scene feature matching and user usage habits to reason about the usage environment data to obtain an accurate target service ID. Compared with a single matching logic, this significantly improves the accuracy of routing judgment and avoids routing conflicts of multiple services under the same AID. Subsequently, the HCE service corresponding to the target service ID can be called to respond to the flashing operation to ensure the accuracy of service routing. Then, user profile samples can be generated based on the flashing results to provide real and reliable training materials for the continuous optimization of the model. That is, the service routing model is optimized by using the user profile samples, so that the model can continuously learn the user's personalized usage preferences and scene change patterns, and realize the dynamic self-evolution of the model.

[0122] In one embodiment, the data acquisition module 210 may include:

[0123] The parameter acquisition submodule is used to collect the target user's current geographical location and current timestamp at the time of flashing.

[0124] The data generation submodule is used to generate usage environment data for the target user based on the current geographical location and the current timestamp.

[0125] In one embodiment, the model prediction module 220 may include:

[0126] The text generation submodule is used to obtain all HCE services under the application identifier and generate scenario description text for each HCE service.

[0127] The database generation submodule is used to construct a vector database of application identifiers based on the description text of each scenario, and to collect the usage environment data of target users for each HCE service in a historical time period to form a personal profile database.

[0128] The model training submodule is used to fine-tune the preset large language model using a vector database and a personal profile database to obtain the service routing model corresponding to the application identifier.

[0129] In one embodiment, the database generation submodule may include:

[0130] The feature recognition unit is used to perform word segmentation on the description text of each scene, obtain the word segmentation results, and identify the core feature words of each description text from the word segmentation results.

[0131] The vector transformation unit is used to convert each core feature word into a scene description vector of a preset dimension, and binds and stores each feature embedding vector with the service ID of the corresponding HCE service to form a vector database corresponding to the application identifier.

[0132] In one embodiment, the service routing model in the model prediction module 220 may include a similarity retrieval network, a contextual reasoning network, and a fusion decision network; the model prediction module 220 may also include:

[0133] The similarity matching submodule is used to generate the current environment vector of the environment data using the similarity retrieval network, and to perform similarity matching between the current environment vector and each scene description vector in the vector database to obtain the scene matching result.

[0134] The Intent Reasoning Submodule is used to input usage environment data into the Contextual Reasoning Network, so that the Contextual Reasoning Network can perform intent reasoning on the usage environment data based on the personal profile database and output the intent reasoning result.

[0135] The fusion decision submodule is used to perform fusion decision-making on scene matching results and intent reasoning results through the fusion decision network to obtain the target service ID.

[0136] In one embodiment, the fusion decision submodule may include:

[0137] The ID judgment unit is used to determine whether the predicted service ID in the scene matching result and the intent reasoning result are consistent using the fusion decision network.

[0138] The first output unit is used to output the predicted service ID as the target service ID through the fusion decision network if the condition is met.

[0139] The second output unit is used to output the predicted service ID of the intent reasoning result as the target service ID through the fusion decision network if the confidence level of the intent reasoning result exceeds a preset threshold.

[0140] The third output unit is used to trigger a pop-up confirmation mode when the predicted service ID in the scene matching result and the intent reasoning result are inconsistent and the confidence level of the intent reasoning result does not exceed a preset threshold, so that the intent reasoning result outputs the target service ID according to the confirmation result of the target user.

[0141] In one embodiment, the sample generation module 230 may include:

[0142] The first generation submodule is used to generate a user profile sample based on the usage environment data and the target service ID when the flashing result is successful.

[0143] The second generation submodule is used to monitor the correction service ID selected by the target user when the flashing result fails, and generate a user profile sample based on the usage environment data, the target service ID, and the correction service ID.

[0144] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the host card emulation service routing method as described in any of the above embodiments.

[0145] In one embodiment, this application also provides a computer device storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the host card emulation service routing method as described in any of the above embodiments.

[0146] Indicatively, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the internal structure of a computer device 300 provided in an embodiment of this application. The computer device 300 can be provided as a server. (Refer to...) Figure 4 The computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by memory 301 for storing instructions, such as application programs, that can be executed by the processing component 302. The application programs stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 302 is configured to execute instructions to perform the host card emulation service routing method of any of the above embodiments.

[0147] The computer device 300 may also include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate on an operating system stored in memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0148] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0149] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0150] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0151] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for simulating service routing on a host card, characterized in that, The method includes: When a target user is detected performing an HCE flashing operation, the user's usage environment data is collected. Determine the service routing model corresponding to the application identifier used by the target user, and use the service routing model to perform dual-path fusion prediction of similarity retrieval and usage context reasoning on the usage environment data to obtain the target service ID; The HCE service corresponding to the target service ID is invoked to respond to the flashing operation, the flashing result is obtained, and a user profile sample is generated based on the flashing result; The service routing model is weighted and optimized using the user profile samples to obtain the latest service routing model for the application identifier.

2. The host card emulation service routing method according to claim 1, characterized in that, The collection of the target user's usage environment data includes: Collect the target user's current geographical location and current timestamp at the time of flashing; The target user's usage environment data is generated based on the current geographical location and the current timestamp.

3. The host card emulation service routing method according to claim 1, characterized in that, The process of constructing the service routing model includes: Obtain all HCE services under the application identifier and generate scenario description text for each HCE service; A vector database of application identifiers is constructed based on the description text of each scenario, and the usage environment data of the target user for each HCE service in a historical time period is collected to form a personal profile database. The preset large language model is fine-tuned and trained using the vector database and the personal profile database to obtain the service routing model corresponding to the application identifier.

4. The host card emulation service routing method according to claim 3, characterized in that, The vector database of application identifiers constructed based on the description text of each scenario includes: The text describing each scene is segmented into words to obtain the segmentation results, and the core feature words of each scene description text are identified from the segmentation results. Each core feature word is converted into a scenario description vector of a preset dimension, and each feature embedding vector is bound and stored with the service ID of the corresponding HCE service to form a vector database corresponding to the application identifier.

5. The host card emulation service routing method according to claim 3, characterized in that, The service routing model includes a similarity retrieval network, a contextual reasoning network, and a fusion decision network. The step of using the service routing model to perform dual-path fusion prediction on the usage environment data through similarity retrieval and usage context reasoning to obtain the target service ID includes: The similarity retrieval network is used to generate the current environment vector of the used environment data, and the current environment vector is matched with each scene description vector in the vector database to obtain the scene matching result; The usage environment data is input into the contextual reasoning network, so that the contextual reasoning network performs intent reasoning on the usage environment data based on the personal profile database and outputs the intent reasoning result; The target service ID is obtained by fusing the scene matching result and the intent reasoning result through the fusion decision network.

6. The host card emulation service routing method according to claim 5, characterized in that, The step of fusing the scene matching result and the intent reasoning result through the fusion decision network to obtain the target service ID includes: The fusion decision network is used to determine whether the prediction service ID in the scene matching result and the intent reasoning result are consistent. If so, the predicted service ID is output through the fusion decision network as the target service ID; If not, when the confidence level of the intent reasoning result exceeds a preset threshold, the predicted service ID of the intent reasoning result is output through the fusion decision network as the target service ID; When the predicted service ID in the scene matching result and the intent reasoning result are inconsistent and the confidence level of the intent reasoning result does not exceed the preset threshold, a pop-up confirmation mode is triggered so that the intent reasoning result outputs the target service ID based on the confirmation result of the target user.

7. The host card emulation service routing method according to claim 1, characterized in that, The step of generating a user profile sample based on the flashing result includes: When the flashing result is successful, a user profile sample is generated based on the usage environment data and the target service ID; When the flashing process fails, the system monitors the correction service ID selected by the target user and generates a user profile sample based on the usage environment data, the target service ID, and the correction service ID.

8. A host card emulation service routing device, characterized in that, include: The data acquisition module is used to collect the target user's usage environment data when the target user performs an HCE flashing operation. The model prediction module is used to determine the service routing model corresponding to the application identifier used by the target user, and to use the service routing model to perform dual-path fusion prediction of similarity retrieval and usage context reasoning on the usage environment data to obtain the target service ID. The sample generation module is used to call the HCE service corresponding to the target service ID to respond to the flashing operation, obtain the flashing result, and generate a user profile sample based on the flashing result; The model optimization module is used to optimize the service routing model with weights using the user profile samples to obtain the latest service routing model for the application identifier.

9. A storage medium, characterized in that: The storage medium stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the host card emulation service routing method as described in any one of claims 1 to 7.

10. A computer device, characterized in that, include: One or more processors, and memory; The memory stores computer-readable instructions that, when executed by the one or more processors, perform the steps of the host card emulation service routing method as described in any one of claims 1 to 7.

Images