Package recommendations: methods, devices, electronic equipment, and storage media

Abstract

This invention discloses a package recommendation method, apparatus, electronic device, and storage medium. The package recommendation method includes: acquiring user data of target users from a telecommunications operator; inputting the user data of the target users from the telecommunications operator and package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target user and each package; the package recommendation model is pre-trained through federated learning based on user data samples of users shared by the telecommunications operator and a third-party platform, and package-related data samples of the telecommunications operator; and recommending packages to the target user based on the matching degree between the target user's user data and each package. This invention can effectively improve the accuracy of package recommendations and enhance the user experience when using packages.

Description

Technical Field

[0001] This invention relates to the field of Internet technology, specifically to a method, apparatus, electronic device, and storage medium for recommending data packages. Background Technology

[0002] To achieve business growth and reduce churn rates, telecom operators have launched a variety of packages, such as various 5G packages. Currently, the main methods for recommending communication packages to users are as follows: 1. Manual recommendations from service centers, where customer service staff recommend suitable personalized packages based on their sales experience; 2. Semi-automatic recommendations, which start from an online perspective, collecting user feedback information such as personal preferences and consumption habits, and then using recommendation models such as collaborative filtering, logistic regression, clustering, and decision trees to recommend suitable packages; 3. Fully automatic recommendations, which utilize the characteristics of existing user data from operators, mining users' historical package information and call / data usage, and then using recommendation models to calculate the user's match with existing packages, thereby recommending the most suitable communication package.

[0003] The following technical shortcomings exist: Manually recommended packages at service centers rely entirely on customer service staff's sales experience, resulting in high costs and low efficiency, as well as subjectivity and incompleteness. Semi-automatic recommendation methods require users to clearly define their needs and actively participate in feedback, making it difficult to obtain sufficient user data. User preference descriptions are inaccurate, and only a small percentage of users benefit, leading to low recommendation effectiveness. Fully automatic recommendation methods can only utilize the operator's user characteristics, failing to provide a more comprehensive user description. Directly using other dimensions of user characteristics inevitably carries the risk of user data privacy leaks. Furthermore, recommendation models based on traditional machine learning algorithms cannot extract deeper features from user data, resulting in low recommendation accuracy. Summary of the Invention

[0004] Based on the problems existing in the prior art, embodiments of the present invention propose a package recommendation method, apparatus, electronic device, and storage medium.

[0005] In a first aspect, embodiments of the present invention provide a method for recommending package deals, including:

[0006] Obtain user data of target users of telecommunications operators;

[0007] The user data of the target users of the telecom operator and the package-related data of each package are input into a pre-trained package recommendation model to obtain the matching degree between the target users and each package. The package recommendation model is pre-trained by federated learning based on user data samples of users shared by the telecom operator and third-party platforms and package-related data samples of the telecom operator.

[0008] Based on the matching degree between the target user's user data and the various packages, a package is recommended to the target user.

[0009] In some examples, before inputting the user data of the target user of the telecom operator and the package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target user and each package, the process includes: training the package recommendation model through federated learning based on user data samples of users shared by the telecom operator and third-party platforms and package-related data samples of the telecom operator.

[0010] In some examples, the user data samples shared by users from telecom operators and third-party platforms, along with the telecom operator's package-related data samples, are used to train the package recommendation model through federated learning, including:

[0011] Obtain the shared users among the aforementioned telecom operators and third-party platforms;

[0012] Public keys are distributed to the telecommunications operators and third-party platforms so that the telecommunications operators and third-party platforms can interact in encrypted form with intermediate results used to compute gradients;

[0013] The telecom operator and the third-party platform each perform calculations based on encrypted gradient values. Meanwhile, the telecom operator calculates the loss based on relevant data samples of the corresponding package.

[0014] The encrypted gradient values ​​and the loss from the telecom operator and the third-party platform are aggregated, and the total gradient value is calculated based on the aggregated results. The total gradient value is then decrypted and sent back to the telecom operator and the third-party platform respectively, so that the telecom operator can update the model parameters of the package recommendation model.

[0015] In some examples, obtaining users shared by the telecom operator and the third-party platform includes:

[0016] Based on encrypted user sample alignment technology, the users shared by the telecom operator and the third-party platform are obtained.

[0017] In some examples, recommending packages to the target user based on the matching degree between the target user's user data and the various packages includes:

[0018] The matching degree between the target user's user data and each of the packages is sorted.

[0019] One or more packages with high matching degree are recommended to the target user.

[0020] In some examples, recommending one or more highly relevant packages to the target user includes:

[0021] A package recommendation list is generated based on the one or more package recommendations, and the package recommendation list is recommended to the target user.

[0022] In some examples, the third-party platform is an e-commerce platform.

[0023] Secondly, embodiments of the present invention provide a package recommendation device, comprising:

[0024] The acquisition module is used to acquire user data of target users of telecommunications operators.

[0025] The matching module is used to input the user data of the target users of the telecom operator and the package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target users and each package. The package recommendation model is pre-trained by federated learning based on user data samples of users shared by the telecom operator and third-party platforms and package-related data samples of the telecom operator.

[0026] The recommendation module is used to recommend packages to the target user based on the matching degree between the target user's user data and the various packages.

[0027] Thirdly, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the package recommendation method as described in the first aspect.

[0028] Fourthly, embodiments of the present invention also provide a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the package recommendation method as described in the first aspect.

[0029] As can be seen from the above technical solutions, the package recommendation method, device, electronic device, and storage medium provided in the embodiments of the present invention, under the condition that the data of each platform is kept on the local data server and does not go out, utilize federated learning technology to jointly establish a deep neural network model on user data from operator platforms and user data from third-party platforms. This not only expands the feature dimensions of operator users while protecting user data privacy, but also uses deep learning in the package recommendation model to mine deeper and more comprehensive user features, thereby improving the accuracy of package recommendations and enhancing the user experience of using packages. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0031] Figure 1 This is a flowchart of a package recommendation method provided in an embodiment of the present invention;

[0032] Figure 2 This is a schematic diagram of federated learning of the model of the package recommendation method provided in another embodiment of the present invention;

[0033] Figure 3 This is a schematic diagram of the two half-model structures of the operator and the e-commerce platform in the package recommendation method provided in another embodiment of the present invention;

[0034] Figure 4 This is a flowchart illustrating the package recommendation process of another embodiment of the present invention.

[0035] Figure 5 This is a structural block diagram of a package recommendation device provided in an embodiment of the present invention;

[0036] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0037] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. The following examples are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0038] The following describes, with reference to the accompanying drawings, a method, apparatus, electronic device, and storage medium for recommending packages according to embodiments of the present invention.

[0039] Figure 1 A flowchart of a package recommendation method according to an embodiment of the present invention is shown. Figure 1 As shown, an embodiment of the present invention provides a package recommendation method, which includes the following:

[0040] S101: Obtain user data of target users of telecommunications operators.

[0041] User data of target users of telecommunications operators, also known as user data of telecommunications operator platforms, includes, but is not limited to: user identity ID, call duration information, data usage, SMS usage, location information, APP preferences, 5G package content, and the matching degree between the user and the corresponding 5G package. For example, user ID number, average monthly call duration, average monthly data usage, average monthly number of SMS messages sent, average monthly cross-province frequency, music APP preferences, video APP preferences, package data allowance, package call duration, package value-added services, package price, and the matching degree between the user and the package.

[0042] S102: Input the user data of the target user of the telecom operator and the package-related data of each package into the pre-trained package recommendation model to obtain the matching degree between the target user and each package. The package recommendation model is pre-trained by federated learning based on the user data samples of users shared by the telecom operator and the third-party platform and the package-related data samples of the telecom operator.

[0043] In one embodiment of the present invention, before inputting the user data of the target user of the telecommunications operator and the package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target user and each package, the method includes: training the package recommendation model through federated learning based on user data samples of users shared by the telecommunications operator and third-party platforms and package-related data samples of the telecommunications operator.

[0044] In this example, the package recommendation model is trained using federated learning based on user data samples shared by the telecom operator and third-party platforms, and package-related data samples from the telecom operator. This includes:

[0045] Obtain the shared users among the aforementioned telecom operators and third-party platforms;

[0046] Public keys are distributed to the telecommunications operators and third-party platforms so that the telecommunications operators and third-party platforms can interact in encrypted form with intermediate results used to compute gradients;

[0047] The telecom operator and the third-party platform each perform calculations based on encrypted gradient values. Meanwhile, the telecom operator calculates the loss based on relevant data samples of the corresponding package.

[0048] The encrypted gradient values ​​and the loss from the telecom operator and the third-party platform are aggregated, and the total gradient value is calculated based on the aggregated results. The total gradient value is then decrypted and sent back to the telecom operator and the third-party platform respectively, so that the telecom operator can update the model parameters of the package recommendation model.

[0049] In the above example, obtaining the shared users between the telecom operator and the third-party platform includes: obtaining the shared users between the telecom operator and the third-party platform based on encrypted user sample alignment technology.

[0050] In the above description, the third-party platform is, for example, an e-commerce platform. The following detailed description will use an e-commerce platform as an example.

[0051] Combination Figure 2 As shown, telecommunications operators and e-commerce platforms upload standardized user data to their local servers. The user data on telecommunications operator platforms includes user ID, call duration, data usage, SMS usage, location information, app preferences, 5G plan details, and the user's compatibility with their corresponding 5G plan. Examples include user ID number, average monthly call duration, average monthly data usage, average monthly number of SMS messages sent, average monthly cross-province frequency, music app preferences, video app preferences, plan data allowance, plan call duration, plan value-added services, plan price, and the user's compatibility with their plan. The user data on e-commerce platforms includes user ID, historical shopping information, and consumption patterns, such as user ID number, average monthly number of purchases, most frequent purchase type, and average monthly spending.

[0052] The local servers of each party perform encrypted sample alignment to obtain a shared user list. Using encryption technologies such as homomorphic encryption and RSA encryption, the encrypted user identity IDs are sent to data fusion center server A. Data fusion center server A calculates the intersection of the encrypted user identity ID data and sends the encrypted shared user identity IDs to user data servers B and C. After local decryption, the shared user identity IDs are intersected with the original data based on the user identity IDs to obtain the final shared user list.

[0053] The data fusion center server distributes initialized deep neural network semi-models to user data server B and user data server C. The semi-model structure obtained by the operator platform's local server B is the core model of this technical solution. Its input layer consists of user data features from the operator platform and 5G package data features, and its output layer is the relevance output obtained from this semi-model. The semi-model structure obtained by the e-commerce platform's local server C has its input layer composed of user data features from the e-commerce platform, and its output layer is the relevance output obtained from this semi-model. The structures of the two semi-models are as follows: Figure 3 As shown, this refers to federated learning.

[0054] The operator platform's local server B and the e-commerce platform's local server C send encrypted intermediate results result_B and result_C to the data fusion center server A, respectively. The operator platform's local server B sends the encrypted user-package matching score match_score to the data fusion center server A.

[0055] Data fusion center server A calculates the gradients gradients gradient_B and gradient_C of each party's semi-model based on result_B, result_C, and match_score, and sends them to the local servers of each platform. The local servers of each platform decrypt the gradients and update their respective models.

[0056] S103: Based on the matching degree between the target user's user data and each package, recommend a package to the target user.

[0057] In one embodiment of the present invention, recommending packages to the target user based on the matching degree between the target user's user data and the various packages includes: sorting the matching degree between the target user's user data and the various packages; and recommending one or more packages with high matching degree to the target user.

[0058] In this example, recommending one or more packages with high matching degree to the target user includes: generating a package recommendation list based on the one or more package recommendations, and recommending the package recommendation list to the target user.

[0059] like Figure 4 As shown, the telecommunications operator platform and the e-commerce platform upload user data to the local server, and the data feature types and standardization methods must be consistent with the data upload module.

[0060] The operator platform's local server B and the e-commerce platform's local server C encrypt and upload the results of their local semi-model calculations to the data fusion center server A.

[0061] The data fusion center server A calculates and summarizes the results, then sends the results back to the operator platform's local server B. After decryption, the match score between the end user and the corresponding 5G package is obtained.

[0062] The matching degree between the user and other 5G plans is calculated using a recommendation model. The matching degrees are then ranked, and the top-ranked plans are selected as the user's 5G plan recommendation list.

[0063] According to the package recommendation method of this invention, under the condition that the data of each platform is kept on the local data server and does not go out, federated learning technology is used to jointly build a deep neural network model on user data from operator platforms and third-party platforms. This not only expands the feature dimensions of operator users while protecting user data privacy, but also uses deep learning in the package recommendation model to mine deeper and more comprehensive user features, thereby improving the accuracy of package recommendations and enhancing the user experience.

[0064] Figure 5A schematic diagram of the package recommendation device provided in an embodiment of the present invention is shown, as follows: Figure 5 As shown, the package recommendation device provided in this embodiment of the invention includes: an acquisition module 510, a matching module 520, and a recommendation module 530, wherein:

[0065] The acquisition module 510 is used to acquire user data of target users of telecommunications operators;

[0066] The matching module 520 is used to input the user data of the target user of the telecom operator and the package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target user and each package. The package recommendation model is pre-trained by federated learning based on the user data samples of users shared by the telecom operator and the third-party platform and the package-related data samples of the telecom operator.

[0067] The recommendation module 530 is used to recommend packages to the target user based on the matching degree between the target user's user data and the various packages.

[0068] According to the package recommendation device of the present invention, under the condition that the data of each platform is kept on the local data server and does not go out, the federated learning technology is used to jointly build a deep neural network model on the user data of the operator platform and the user data of the third-party platform. This not only expands the feature dimensions of the operator users while protecting the privacy of user data, but also uses deep learning in the package recommendation model to mine deeper and more comprehensive user features, thereby improving the accuracy of package recommendation and enhancing the user's experience of using the package.

[0069] It should be noted that the specific implementation of the package recommendation device in this embodiment of the invention is similar to the specific implementation of the package recommendation method in this embodiment of the invention. Please refer to the description in the method section for details. In order to reduce redundancy, the details will not be repeated here.

[0070] Based on the same inventive concept, another embodiment of the present invention provides an electronic device, see [link to previous document]. Figure 6 The electronic device specifically includes the following components: processor 401, memory 402, communication interface 403, and communication bus 404;

[0071] The processor 401, memory 402, and communication interface 403 communicate with each other through the communication bus 404; the communication interface 403 is used to realize information transmission between the devices.

[0072] The processor 401 is used to call the computer program in the memory 402. When the processor executes the computer program, it implements all the steps of the above-mentioned package recommendation method. For example, when the processor executes the computer program, it implements the following steps: obtaining user data of the target user of the telecommunications operator; inputting the user data of the target user of the telecommunications operator and the package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target user and each package, wherein the package recommendation model is pre-trained through federated learning based on user data samples of users shared by the telecommunications operator and third-party platforms and package-related data samples of the telecommunications operator; and recommending packages to the target user based on the matching degree between the user data of the target user and each package.

[0073] Based on the same inventive concept, another embodiment of the present invention provides a non-transitory computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements all the steps of the above-described package recommendation method. For example, when the processor executes the computer program, it implements the following steps: acquiring user data of target users of a telecommunications operator; inputting the user data of the target users of the telecommunications operator and package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target user and each package, wherein the package recommendation model is pre-trained through federated learning based on user data samples of users shared by the telecommunications operator and third-party platforms and package-related data samples of the telecommunications operator; and recommending packages to the target user based on the matching degree between the target user's user data and each package.

[0074] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0075] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the embodiments of the present invention according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0076] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the indicator monitoring method described in various embodiments or some parts of embodiments.

[0077] Furthermore, in this invention, terms such as "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0078] Furthermore, in this invention, relational terms such as "first" and "second" are used merely 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. Moreover, 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 limitation, 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.

[0079] Furthermore, in the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for recommending meal packages, characterized in that, include: Obtain user data of target users of telecommunications operators; The user data of the target users of the telecommunications operator and the package-related data of each package are input into a pre-trained package recommendation model to obtain the matching degree between the target users and each package. Based on the matching degree between the target user's user data and the various packages, a package is recommended to the target user; The package recommendation model is determined based on the following steps: Obtain the shared users among the aforementioned telecom operators and third-party platforms; The public key is distributed to the telecom operators and third-party platforms. The local servers of the operator platforms and the third-party platforms send encrypted intermediate results to the data fusion center server. The local servers of the operator platforms send encrypted user and package matching degrees to the data fusion center server. The data fusion center server calculates gradients based on the intermediate results and matching degrees and distributes them to the local servers of each platform. The telecom operator and the third-party platform each perform calculations based on encrypted gradient values. Meanwhile, the telecom operator calculates the loss based on relevant data samples of the corresponding package. The encrypted gradient values ​​and the loss from the telecom operator and the third-party platform are aggregated, and the total gradient value is calculated based on the aggregated results. The total gradient value is then decrypted and sent back to the telecom operator and the third-party platform respectively, so that the telecom operator can update the model parameters of the package recommendation model.

2. The package recommendation method according to claim 1, characterized in that, Before inputting the user data of the target users of the telecom operator and the package-related data of each package into the pre-trained package recommendation model to obtain the matching degree between the target users and each package, the process includes: training the package recommendation model through federated learning based on user data samples of users shared by the telecom operator and third-party platforms and package-related data samples of the telecom operator.

3. The package recommendation method according to claim 1, characterized in that, The method of obtaining users shared by the telecom operator and the third-party platform includes: Based on encrypted user sample alignment technology, the users shared by the telecom operator and the third-party platform are obtained.

4. The package recommendation method according to claim 1, characterized in that, The step of recommending packages to the target user based on the matching degree between the target user's user data and the various packages includes: The matching degree between the target user's user data and each of the packages is sorted. One or more packages with high matching degree are recommended to the target user.

5. The package recommendation method according to claim 4, characterized in that, The step of recommending one or more packages with high matching degree to the target user includes: A package recommendation list is generated based on the one or more package recommendations, and the package recommendation list is recommended to the target user.

6. The package recommendation method according to any one of claims 1-5, characterized in that, The third-party platform mentioned is an e-commerce platform.

7. A package recommendation device, characterized in that, include: The acquisition module is used to acquire user data of target users of telecommunications operators. The matching module is used to input the user data of the target users of the telecom operator and the package-related data of each package into a pre-trained package recommendation model to obtain the matching degree between the target users and each package. The package recommendation model is pre-trained by federated learning based on user data samples of users shared by the telecom operator and third-party platforms and package-related data samples of the telecom operator. The recommendation module is used to recommend packages to the target user based on the matching degree between the target user's user data and the various packages; The package recommendation model is determined based on the following steps: Obtain the shared users among the aforementioned telecom operators and third-party platforms; The public key is distributed to the telecom operators and third-party platforms. The local servers of the operator platforms and the third-party platforms send encrypted intermediate results to the data fusion center server. The local servers of the operator platforms send encrypted user and package matching degrees to the data fusion center server. The data fusion center server calculates gradients based on the intermediate results and matching degrees and distributes them to the local servers of each platform. The telecom operator and the third-party platform each perform calculations based on encrypted gradient values. Meanwhile, the telecom operator calculates the loss based on relevant data samples of the corresponding package. The encrypted gradient values ​​and the loss from the telecom operator and the third-party platform are aggregated, and the total gradient value is calculated based on the aggregated results. The total gradient value is then decrypted and sent back to the telecom operator and the third-party platform respectively, so that the telecom operator can update the model parameters of the package recommendation model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the package recommendation method according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the package recommendation method according to any one of claims 1 to 6.

Images