Information recommendation method and device, computer program product and electronic equipment

Abstract

The embodiment of the invention relates to an information recommendation method and device, a computer program product and electronic equipment, and relates to the technical field of computers.The method comprises the steps that sample user data is obtained from a target environment, model parameters of a recommendation model are adjusted based on the sample user data, and a trained recommendation model is generated; performing zero-knowledge certification on the trained recommendation model to determine a certification result; and under the condition that the certification result is that verification is passed, performing feature fitting on user data of a target user based on the trained recommendation model, and determining recommendation information for the target user. According to the method and the device, the recommendation information for the target user can be accurately determined, and precise marketing is realized.

Description

Background Art

[0002] As the market gradually becomes saturated, the growth of operators' users is gradually slowing down. In order to attract customers, operators need to invest more resources and energy. Operators will excessively collect user data to improve service quality, thereby ensuring user praise and retention rates.

[0003] The above method involves user privacy issues in the process of collecting user data and cannot be used publicly, resulting in low accuracy of recommendations and reduced service quality. Summary of the invention

[0004] The purpose of the present disclosure is to provide an information recommendation method, an information recommendation device, a computer program product, and an electronic device, thereby overcoming, at least to a certain extent, the problem of low accuracy of recommendation results due to the possible leakage of user privacy caused by the limitations and defects of related technologies.

[0005] According to one aspect of the present disclosure, there is provided an information recommendation method, comprising: acquiring sample user data from a target environment, and adjusting model parameters of a recommendation model based on the sample user data to generate a trained recommendation model; performing zero-knowledge proof on the trained recommendation model to determine a proof result; when the proof result is verified to be passed, performing feature fitting on the user data of the target user based on the trained recommendation model to determine recommended information for the target user.

[0006] In an exemplary embodiment of the present disclosure, the model parameters of the recommendation model are adjusted based on the sample user data to generate a trained recommendation model, including: classifying the sample user data through a support vector machine to determine multiple categories; adjusting the model parameters of the recommendation model according to the sample user data of multiple categories to generate the trained recommendation model.

[0007] In an exemplary embodiment of the present disclosure, the model parameters of the recommendation model are adjusted according to sample user data of multiple categories to generate the trained recommendation model, including: extracting features of the sample user data of multiple categories to determine multiple reference user features; screening the user features from the multiple reference user features according to attribute parameters of the reference user features; adjusting the model parameters of the recommendation model based on the user features to generate the trained recommendation model; wherein the attribute parameters include one or more of contribution and correlation.

[0008] In an exemplary embodiment of the present disclosure, the zero-knowledge proof of the trained recommendation model to determine the proof result includes: generating a proof for the sample user data, and verifying the proof to obtain the proof result; wherein the proof is used to prove that the data for training the recommendation model is the sample user data in the target environment.

[0009] In an exemplary embodiment of the present disclosure, generating a proof for the sample user data includes: determining statement information for the sample user data; converting the statement information into a circuit based on a zero-knowledge proof algorithm, defining public data and private data in the circuit, and compiling the circuit to generate a witness; and selecting target parameters through the witness to generate the proof.

[0010] In an exemplary embodiment of the present disclosure, the verification of the proof to obtain the proof result includes: converting the proof based on the zero-knowledge proof algorithm to obtain a proof equation; solving the proof equation, and determining the proof result according to the solution result of the proof equation.

[0011] In an exemplary embodiment of the present disclosure, determining the proof result according to the solution result of the proof equation includes: if the solution result of the proof equation is correct, determining that the proof result is verified; if the solution result of the proof equation is wrong, determining that the proof result is not verified.

[0012] According to one aspect of the present disclosure, there is provided an information recommendation device, comprising: a model training module, used to obtain sample user data from a target environment, and adjust model parameters of a recommendation model based on the sample user data to generate a trained recommendation model; a zero-knowledge proof module, used to perform zero-knowledge proof on the trained recommendation model to determine a proof result; and a recommendation information determination module, used to perform feature fitting on user data of a target user based on the trained recommendation model to determine recommendation information for the target user when the proof result is verified.

[0013] According to one aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the information recommendation method described above is implemented.

[0014] According to one aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any one of the above-mentioned information recommendation methods by executing the executable instructions.

[0015] In the technical solution provided in the embodiment of the present disclosure, on the one hand, the recommendation model is trained by collecting sample user data in the target environment, and then the data for training the recommendation model is determined to be the sample user data in the target environment through zero-knowledge proof. Since the data inside the target environment is used for model training and zero-knowledge proof is used for verification, the sample user data can be safely used for model training while ensuring user privacy, which improves the comprehensiveness of the data used in training the model and improves the accuracy of the recommendation model. On the other hand, since the accuracy of the recommendation model can be improved, precision marketing for target users can be achieved based on the recommendation model, the accuracy of the recommendation information can be improved, the service quality can be improved, and the user retention rate can be guaranteed.

[0016] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0018] Figure 1 The following schematically shows a flow chart of an information recommendation method in an embodiment of the present disclosure.

[0019] Figure 2 The flowchart of training the recommendation model in the embodiment of the present disclosure is schematically shown.

[0020] Figure 3 The flowchart of generating a certificate is schematically shown.

[0021] Figure 4 A flowchart schematically illustrates a data security marketing method based on zero-knowledge proof.

[0022] Figure 5 The following is a schematic diagram of a block diagram of an information recommendation device in an embodiment of the present disclosure.

[0023] Figure 6 The block diagram of the electronic device according to the embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0024] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0025] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0026] As the market gradually becomes saturated, the growth of operators' users is gradually slowing down. In order to win customers, operators need to invest more resources and energy. At the same time, operators will excessively collect user information data to improve service quality, thereby ensuring user praise and retention rates. When collecting user information data, the marketing data currently designed on the Internet involves user privacy issues and cannot be extracted and used at will, so the privacy is poor. In addition, since user data cannot be extracted and used at will, the amount of data that can be extracted is small, making it difficult to recommend information suitable for users, resulting in poor service quality and difficulty in ensuring user retention.

[0027] In order to solve the above technical problems, an information recommendation method is provided in an embodiment of the present disclosure, which is used to provide users with accurate recommendation information without leaking user privacy, realize precision marketing, thereby improving the quality of operation services and ensuring the user's favorable rate and retention rate. The information recommendation method can be applied to various types of recommendation scenarios, such as product recommendation, marketing recommendation, etc.

[0028] Figure 1 A flowchart of the information recommendation method is schematically shown in FIG. Figure 1 As shown in , the information recommendation method mainly includes the following steps:

[0029] In step S110, sample user data is obtained from the target environment, and model parameters of the recommendation model are adjusted based on the sample user data to generate a trained recommendation model;

[0030] In step S120, a zero-knowledge proof is performed on the trained recommendation model to determine a proof result;

[0031] In step S130, when the certification result is passed, feature fitting is performed on the user data of the target user based on the trained recommendation model to determine the recommendation information for the target user.

[0032] The technical solution provided in the disclosed embodiment collects sample user data in the target environment to train the recommendation model, and then determines that the sample user data in the target environment is used to train the recommendation model through zero-knowledge proof. In order to solve the problem that the data cannot be used publicly, the privacy of the data is guaranteed through zero-knowledge proof. In addition, due to the zero-knowledge proof, the sample user data can be safely used for model training while ensuring the privacy of the user, ensuring the privacy and confidentiality of the data when training and evaluating the machine learning model on sensitive data. The accuracy of the recommendation model is improved, and precision marketing for target users can be achieved. On this basis, since the accuracy of the recommended information can be improved, the service quality can be improved and the user retention rate can be guaranteed.

[0033] Next, refer to Figure 1 The information recommendation method in the embodiment of the present disclosure is specifically described in detail.

[0034] In step S110, sample user data is obtained from the target environment, and model parameters of the recommendation model are adjusted based on the sample user data to generate a trained recommendation model.

[0035] In the embodiment of the present disclosure, the target environment may be a network environment within a target organization, which may be a company or other organization. Different target organizations may have different corresponding target environments. For example, the target environment may be the existing network environment of Company A.

[0036] For example, sample user data can be obtained from the target environment, and the sample user data can be data that plays a key role in user services. When the target environment is the operator's existing network environment, the sample user data can include one or more of user personal information, daily traffic usage, package usage, package subscription, and application usage habits.

[0037] It should be noted that the sample user data collected from the target environment can only be used within the target organization and cannot be used externally to improve data privacy.

[0038] After collecting sample user data from the target environment, the model parameters of the recommendation model can be adjusted based on the sample user data to generate a trained recommendation model. The recommendation model can be trained in the target environment.

[0039] Figure 2 The flowchart of training the recommendation model is shown schematically in Figure 2 As shown in , it mainly includes the following steps:

[0040] In step S210, the sample user data is classified by a support vector machine to determine a plurality of categories;

[0041] In step S220, feature extraction is performed on the sample user data of multiple categories to determine multiple reference user features;

[0042] In step S230, the user feature is selected from a plurality of reference user features according to the attribute parameters of the reference user feature;

[0043] In step S240, the model parameters of the recommendation model are adjusted based on the user characteristics to generate a trained recommendation model.

[0044] In the disclosed embodiment, a support vector machine (SVM) is a type of generalized linear classifier that performs binary classification on data in a supervised learning manner, and its decision boundary is determined by solving a maximum margin hyperplane for learning samples.

[0045] The support vector machine SVM finds a hyperplane that can accurately distinguish between two types of samples. This hyperplane is called the optimal classification hyperplane. For example, a kernel function can be selected to map the sample data from the original space to a higher-dimensional feature space so that the sample data is linearly separable in the feature space. The SVM model is further trained using the training data set. In this process, the SVM aims to find a hyperplane that can maximize the interval between two different categories. This hyperplane can divide the data points into two different categories. The trained SVM model is applied to the classification prediction of the sample user data to determine multiple categories.

[0046] Furthermore, the model parameters of the recommendation model can be adjusted based on the sample user data of each category, so as to determine a trained recommendation model. The recommendation model can be any suitable machine learning model as long as it can realize the recommendation function.

[0047] In the process of training the recommendation model based on multiple types of sample user data, the sample user data can first be cleaned and formatted to ensure the quality and consistency of the data. Further, feature extraction can be performed on the sample user data of multiple categories to obtain multiple reference user features.

[0048] In marketing analysis for determining recommendation information for users, the reference user feature may be based on the frequency of occurrence of words or phrases, such as word frequency or TF-IDF (Term Frequency_Inverse Document Frequency). Alternatively, the reference user feature may also be based on semantic information of words, such as word vectors.

[0049] After determining the reference user features, multiple reference user features can be screened. Exemplarily, user features for training a recommendation model can be screened out from multiple reference user features based on the attribute parameters of the reference user features. The attribute parameters may be one or more of contribution and relevance. In some embodiments, the score of the reference user feature can be determined based on one or more of the contribution and relevance, and the user feature can be determined in descending order according to the score. When determining the score of the reference user feature based on the contribution and relevance, the weights corresponding to the contribution and relevance can be determined, and then the weighted summation based on the corresponding weights and the corresponding scores is performed to obtain the score of each reference user feature. Since the reference user features can be screened by one or more of the contribution and relevance, unnecessary features can be reduced and the accuracy of the user features used to train the recommendation model can be improved.

[0050] On this basis, the model parameters of the recommendation model can be updated according to the screened user characteristics to generate a trained recommendation model. Since the recommendation model can be trained according to the screened user characteristics, the accuracy and efficiency of the recommendation model can be improved.

[0051] Next, continue to refer to Figure 1 As shown in , in step S120, zero-knowledge proof is performed on the trained recommendation model to determine the proof result.

[0052] In the disclosed embodiment, since the recommendation model is trained in the target environment, when the trained recommendation model is used for reasoning, it is necessary to prove that the recommendation model is trained based on the data of the target environment, that is, to prove that the data used to train the recommendation model is the sample user data in the target environment. In addition, the data features cannot be leaked during the proof process to ensure privacy.

[0053] Zero-knowledge proof refers to a cryptographic tool that allows two untrusting communicating parties to prove the validity of a proposition without revealing any additional information. Zero-knowledge proof allows the prover to prove to the verifier that a statement is true without revealing any additional information other than that the statement is true.

[0054] The zero-knowledge proof of the trained recommendation model mainly includes the process that the prover generates a proof for the sample user data, and sends the proof to the verifier so that the verifier verifies the proof for the sample user data. In the disclosed embodiment, the generated proof is used to prove that the data for training the recommendation model is the sample user data in the target environment, that is, to prove that the sample user data for training the recommendation model comes from the target environment. For example, it is proved that the data used to train the recommendation model comes from the existing network environment.

[0055] When generating a proof for sample user data, the statement information for the sample user data may be converted into a circuit, a witness may be generated based on the circuit, and then a proof may be generated based on the witness. Figure 3 As shown in , it mainly includes the following steps:

[0056] In step S310, statement information for the sample user data is determined.

[0057] In this step, the statement information may be represented by code information. The statement information may be used to represent information associated with the sample user data and the proof. For example, when the proof is used to prove that the data for training the recommendation model is the sample user data in the target environment, the statement information may be code information used to represent that the sample user data was collected from the XX network element in the target environment at XXX time.

[0058] In step S320, the statement information is converted into a circuit based on a zero-knowledge proof algorithm, public data and private data are defined in the circuit, and the circuit is compiled to generate a witness.

[0059] In this step, the zero-knowledge proof algorithm can be a zkp-snark algorithm. The core of the algorithm is to convert the description information into a polynomial form and prove it through homomorphic encryption technology. The specific steps include: converting the statement information into an arithmetic gate circuit, translating the circuit into a matrix formula, and further converting the matrix formula into a polynomial, which should be divisible by a specific minimum polynomial. The last step uses homomorphic encryption to project the statement in the third step into the encrypted space, so that the prover can verify the mirror statement therein.

[0060] The zero-knowledge proof algorithm can convert the declarative information represented by the previous code into a circuit. Among them, the declarative information can be converted into a circuit through the Circom compiler. The circuit can represent the integrity conditions that the recommended model running data needs to meet without displaying the specific content of the data, ensuring the privacy of the data. Based on this, a suitable zkp-snark algorithm can be selected to generate proofs using cryptographic hash functions, elliptic curve encryption and other methods.

[0061] Before providing proof to the verifier, it is necessary to calculate the witness of all signals in the circuit so as to compile the circuit to generate a witness. For example, public data and private data can be defined in the circuit, the circuit can be compiled, and witnesses of the public data and private data can be generated. For example, the sparkjs command can be used to calculate the witness.

[0062] In step S330, the target parameters are selected through the witness to generate the proof.

[0063] In this step, after generating a witness of the statement information, you can select a target parameter, which can be the data in the circuit, and select it according to actual needs to generate a proof. In addition, you can also generate a public file, which can be a subset of the content corresponding to the witness, which only contains the value of the publicly available data in the circuit.

[0064] After obtaining the proof, the generated proof, the zero-knowledge proof algorithm used, and the defined public data can be sent to the verifier. The verifier can verify the received proof using the defined public data and the zero-knowledge proof algorithm used to obtain the proof result. Among them, the integrity, correctness, and zero-knowledge of the proof need to be verified. After receiving the proof, the verifier verifies it according to the zero-knowledge proof protocol used. The zero-knowledge proof protocol can be a zero-knowledge proof algorithm. The verifier may perform a series of calculations or checks to ensure the validity and zero-knowledge of the proof. If the verification is successful, the verifier will accept the prover's assertion; otherwise, it will refuse to accept it.

[0065] Exemplarily, the verifier uses the snarkjs command to verify the received proof. In some embodiments, verifying the received proof may include the following steps: converting the received proof based on a zero-knowledge proof algorithm to convert the statement information into a proof equation; further solving the proof equation, and determining the proof result according to the solution result of the proof equation. The proof equation refers to a mathematical equation. After obtaining the proof equation, the proof equation can be solved to obtain the solution result of the proof equation. If the solution result of the proof equation is correct, it can be determined that the proof result is verified, that is, the declared statement information is correct. Therefore, the data used to train the recommendation model is the sample user data in the target environment. If the solution result of the proof equation is wrong, it can be determined that the proof result is not verified, that is, the declared statement information is wrong, so the data used to train the recommendation model does not belong to the sample user data in the target environment.

[0066] For example, the proof can be converted into Equation 1. Equation 1 can be further solved to determine the solution of Equation 1. If the solution of Equation 1 is correct, the proof result can be considered to be verified. If the solution of Equation 1 is wrong, the proof result can be considered to be verified. When the proof result is verified, it can be determined that the statement information of the declaration is correct, and the data used to train the recommendation model comes from the target environment.

[0067] Continue to refer Figure 1 As shown in , in step S130, when the certification result is verified, feature fitting is performed on the user data of the target user based on the trained recommendation model to determine the recommendation information for the target user.

[0068] In the disclosed embodiment, if the proof result is verification passed, it can be considered that the data used to train the recommendation model is sample user data in the target environment. Based on this, the trained recommendation model can be deployed to the target environment to perform reasoning based on the trained recommendation model. Obtain the user data of the target user, perform feature fitting on the user data of the target user, and determine the recommendation information for the target user. Among them, the target user can be every user of any operator, and the target user is of the same type as the user corresponding to the sample user data in the target environment. User data may include one or more of user personal information, daily traffic usage, package usage, package subscription, and application usage habits.

[0069] After obtaining the user data of the target user, feature extraction can be performed on the user data to determine multiple intermediate user features; further, user features can be screened out from the multiple intermediate user features based on the attribute parameters of the intermediate user features. The intermediate user features can be based on the frequency of occurrence of words or phrases, such as word frequency or TF-IDF (TermFrequency_Inverse Document Frequency). Alternatively, the intermediate user features can also be based on semantic information of words, such as word vectors.

[0070] After determining the plurality of intermediate user features, the plurality of intermediate user features may be screened. Exemplarily, user features may be screened from the plurality of intermediate user features based on attribute parameters of the intermediate user features. The attribute parameters may be one or more of contribution and correlation.

[0071] Furthermore, the user features can be fitted based on the trained recommendation model to obtain recommendation information for the target user. The trained recommendation model here refers to a trained recommendation model that has been proven by zero knowledge.

[0072] The recommendation information may be determined based on the recommendation scenario. The recommendation scenario may be an operator recommendation scenario, a product recommendation scenario on a shopping platform, or a multimedia recommendation scenario on a playback platform. When the recommendation scenario is an operator recommendation scenario, the recommendation information may be, for example, a marketing package for a target user.

[0073] It should be added that after determining the recommendation information for the target user, the sample user data in the target environment can be updated in real time according to the user data of the target user, and the model parameters of the recommendation model can be readjusted based on the updated sample user data, thereby realizing real-time training of the recommendation model and improving the accuracy and reliability of the recommendation model, so that the recommendation model can better adapt to the marketing needs of the operator and realize precision marketing.

[0074] Figure 4 The flowchart of the data security marketing method based on zero-knowledge proof is schematically shown in Figure 2. Combining zero-knowledge proof and machine learning to create a system capable of precision marketing can analyze the exact needs of users and complete precision marketing without leaking user privacy, thereby improving the service quality of operations and ensuring the user's praise rate and retention rate. Figure 4 As shown in , it mainly includes data acquisition module, ZKP integration module and model deployment module, among which:

[0075] In the data collection module, key data of users of the target organization can be collected from the existing network environment. The key data of users includes one or more of user personal information, daily traffic usage, package subscription, and app usage habits. For example, key data can be obtained from IMS as sample user data. IMS (IP Multimedia Subsystem) is a new form of multimedia service that can meet the needs of terminal customers for newer and more diversified multimedia services.

[0076] Next, the user key data can be preprocessed, and the data preprocessing may include data cleaning and data formatting to ensure data consistency. Feature extraction is performed on the preprocessed user key data to obtain reference user features. Each reference user feature can be further screened based on one or more of contribution and correlation to obtain user features for training the recommendation model. By screening each reference user feature by one or more of contribution and correlation, unnecessary features can be reduced, thereby improving the accuracy of the features, and improving the accuracy and efficiency of model training. Based on this, the recommendation model can be trained according to the user features to obtain a trained recommendation model. Exemplarily, the support vector machine SVM algorithm can be used to train the labeled data to obtain a trained recommendation model.

[0077] ZKP integration module: After obtaining the trained recommendation model, you can perform ZKP integration on the trained recommendation model to achieve zero-knowledge proof. Figure 4 As shown in , exemplarily, a zero-knowledge proof algorithm is selected, for example, a zkp-snark algorithm can be selected. Next, the defined statement information can be compiled into a circuit. A proof is generated based on the public data and private data in the circuit. The proof is used to verify that the data for training the recommendation model is sample user data in the target environment. The proof is further verified to obtain a proof result. When verifying, the proof can be converted into a proof equation, and the proof equation is solved to obtain a solution result of the proof equation. Then, the proof result is obtained based on whether the solution result is correct.

[0078] During the model training process, zero-knowledge proof ZKP is used to verify the integrity of the data during model training without leaking the specific data value, and zero-knowledge proof is used to ensure the privacy of the data.

[0079] Model deployment module: The trained recommendation model with zero-knowledge proof can be deployed, and the user data of the target users can be obtained. The user data is then input into the trained recommendation model to determine the recommendation information for the target users and achieve precision marketing.

[0080] Compared with the prior art, the present disclosure addresses the problem that data cannot be used publicly. The data is used while ensuring the privacy of the data through zero-knowledge proof, and the SVM model is used to achieve precision marketing and improve service quality, thereby ensuring the user's praise rate and retention rate.

[0081] The technical solution provided in the disclosed embodiment collects sample user data in the target environment to train the recommendation model, and then determines that the sample user data in the target environment is used to train the recommendation model through zero-knowledge proof. In order to solve the problem that the data cannot be used publicly, the privacy of the data is guaranteed through zero-knowledge proof. In addition, due to the zero-knowledge proof, the sample user data can be safely used for model training while ensuring the privacy of the user, ensuring the privacy and confidentiality of the data when training and evaluating the machine learning model on sensitive data. The accuracy of the recommendation model is improved, and precision marketing for target users can be achieved. On this basis, since the accuracy of the recommended information can be improved, the service quality can be improved and the user retention rate can be guaranteed.

[0082] The present disclosure also provides an information recommendation device. Figure 5 As shown, the information recommendation device 500 mainly includes the following modules:

[0083] A model training module 501 is used to obtain sample user data from a target environment, and adjust model parameters of a recommendation model based on the sample user data to generate a trained recommendation model;

[0084] A zero-knowledge proof module 502 is used to perform zero-knowledge proof on the trained recommendation model to determine the proof result;

[0085] The recommendation information determination module 503 is used to perform feature fitting on the user data of the target user based on the trained recommendation model to determine the recommendation information for the target user when the certification result is verified.

[0086] In an exemplary embodiment of the present disclosure, the model parameters of the recommendation model are adjusted based on the sample user data to generate a trained recommendation model, including: classifying the sample user data through a support vector machine to determine multiple categories; adjusting the model parameters of the recommendation model according to the sample user data of multiple categories to generate the trained recommendation model.

[0087] In an exemplary embodiment of the present disclosure, the model parameters of the recommendation model are adjusted according to sample user data of multiple categories to generate the trained recommendation model, including: extracting features of the sample user data of multiple categories to determine multiple reference user features; screening the user features from the multiple reference user features according to attribute parameters of the reference user features; adjusting the model parameters of the recommendation model based on the user features to generate the trained recommendation model; wherein the attribute parameters include one or more of contribution and correlation.

[0088] In an exemplary embodiment of the present disclosure, the zero-knowledge proof of the trained recommendation model to determine the proof result includes: generating a proof for the sample user data, and verifying the proof to obtain the proof result; wherein the proof is used to prove that the data for training the recommendation model is the sample user data in the target environment.

[0089] In an exemplary embodiment of the present disclosure, generating a proof for the sample user data includes: determining statement information for the sample user data; converting the statement information into a circuit based on a zero-knowledge proof algorithm, defining public data and private data in the circuit, and compiling the circuit to generate a witness; and selecting target parameters through the witness to generate the proof.

[0090] In an exemplary embodiment of the present disclosure, the verification of the proof to obtain the proof result includes: converting the proof based on the zero-knowledge proof algorithm to obtain a proof equation; solving the proof equation, and determining the proof result according to the solution result of the proof equation.

[0091] In an exemplary embodiment of the present disclosure, determining the proof result according to the solution result of the proof equation includes: if the solution result of the proof equation is correct, determining that the proof result is verified; if the solution result of the proof equation is wrong, determining that the proof result is not verified.

[0092] It should be noted that the specific details of each module in the above information recommendation device have been described in detail in the corresponding information recommendation method, so they will not be repeated here.

[0093] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0094] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0095] In an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

[0096] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0097] Refer to the following Figure 6 The electronic device 600 according to this embodiment of the present disclosure is described. Figure 6 The electronic device 600 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0098] like Figure 6 As shown, the electronic device 600 is in the form of a general computing device. The components of the electronic device 600 may include, but are not limited to: the at least one processing unit 610, the at least one storage unit 620, a bus 630 connecting different system components (including the storage unit 620 and the processing unit 610), and a display unit 640.

[0099] The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 610 can perform the following steps: Figure 1 Follow the steps shown in .

[0100] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0101] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0102] Bus 630 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0103] The electronic device 600 may also communicate with one or more external devices 700 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or communicate with any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 650. Furthermore, the electronic device 600 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 660. As shown, the network adapter 660 communicates with other modules of the electronic device 600 via a bus 630. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0104] It should be noted that, in some embodiments of the present disclosure, a computer program product is also provided. The computer program product includes a computer program. When the computer program is executed by a processor, the above method is implemented.

[0105] In one embodiment, a computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing a computer program. The readable storage medium may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, and other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk (HDD), solid-state drive (SSD), and the like. Exemplarily, a computer program product may be implemented as a non-volatile storage medium storing a computer program, such as a read-only memory, a NAND flash memory (Nand Flash), and the like. In one embodiment, a computer program product may be an intangible product containing a computer program. Exemplarily, a computer program product may be implemented as a virtual digital product, such as a digital file such as an executable file, an installation package, etc. storing a computer program.

[0106] The code of the computer program can be written in one or more programming languages. Programming languages ​​such as C language, Java, C++, etc. The program code can be executed entirely on the user computing device, or partially on the user computing device, or as a separate software package, or partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, such as a local area network (LAN), a wide area network (WAN), etc., or can be connected to an external computing device (e.g., an Internet connection provided by an operator).

[0107] The computer program may be carried or transmitted via electrical, magnetic, optical, electromagnetic, infrared, or other signals. The electronic device may convert the signal carrying the computer program into a digital signal, thereby running the computer program. When the computer program is run on an electronic device, its code is used to enable the electronic device to execute (more specifically, the processor of the electronic device may execute) the method steps of various exemplary embodiments of the present disclosure.

[0108] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0109] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0110] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0111] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing what is disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and embodiments are to be considered merely as exemplary, and the true scope and spirit of the present disclosure are indicated by the claims.

[0112] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. An information recommendation method, characterized in that: include: Acquire sample user data from the target environment, and adjust model parameters of the recommendation model based on the sample user data to generate a trained recommendation model; Performing zero-knowledge proof on the trained recommendation model to determine the proof result; When the certification result is that the verification is passed, feature fitting is performed on the user data of the target user based on the trained recommendation model to determine the recommendation information for the target user.

2. The information recommendation method according to claim 1, characterized in that: The adjusting the model parameters of the recommendation model based on the sample user data to generate a trained recommendation model includes: Classifying the sample user data by a support vector machine to determine a plurality of categories; The model parameters of the recommendation model are adjusted according to sample user data of multiple categories to generate the trained recommendation model.

3. The information recommendation method according to claim 2, characterized in that: The step of adjusting the model parameters of the recommendation model according to the sample user data of multiple categories to generate the trained recommendation model includes: Performing feature extraction on the sample user data of multiple categories to determine multiple reference user features; Filtering the user feature from a plurality of reference user features according to the attribute parameters of the reference user feature; Adjusting the model parameters of the recommendation model based on the user characteristics to generate a trained recommendation model; The attribute parameters include one or more of contribution and correlation.

4. The information recommendation method according to claim 1, characterized in that: The performing zero-knowledge proof on the trained recommendation model to determine the proof result includes: Generating a certificate for the sample user data, and verifying the certificate to obtain the certificate result; The proof is used to prove that the data used to train the recommendation model is the sample user data in the target environment.

5. The information recommendation method according to claim 4, characterized in that: The generating of the certificate for the sample user data includes: Determining statement information for the sample user data; Converting the statement information into a circuit based on a zero-knowledge proof algorithm, defining public data and private data in the circuit, and compiling the circuit to generate a witness; The proof is generated by selecting target parameters through the witness.

6. The information recommendation method according to claim 4, characterized in that: The verifying the proof to obtain the proof result includes: Converting the proof based on the zero-knowledge proof algorithm to obtain a proof equation; The proof equation is solved, and the proof result is determined according to the solution result of the proof equation.

7. The information recommendation method according to claim 4, characterized in that: Determining the proof result according to the solution result of the proof equation includes: If the solution result of the proof equation is correct, determining that the proof result is verified; If the solution result of the proof equation is wrong, it is determined that the proof result has failed verification.

8. An information recommendation device, characterized in that: include: A model training module, used to obtain sample user data from a target environment, and adjust model parameters of a recommendation model based on the sample user data to generate a trained recommendation model; A zero-knowledge proof module, used to perform zero-knowledge proof on the trained recommendation model to determine the proof result; The recommendation information determination module is used to perform feature fitting on the user data of the target user based on the trained recommendation model to determine the recommendation information for the target user when the certification result is verified.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the information recommendation method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the information recommendation method according to any one of claims 1 to 7 by executing the executable instructions.

Images