Product recommendation method, device, computer equipment and medium based on federated learning

Abstract

The embodiment of the present application belongs to the field of artificial intelligence, and relates to a product recommendation method based on federated learning, including establishing a federated learning collaborative filtering recommendation model to be trained by combining multiple participants, respectively obtaining the first user features corresponding to the local data of all participants and the characteristic attributes of the products in the product database, inputting the first user features and the characteristic attributes into the federated learning collaborative filtering recommendation model to be trained, and training the similarity between users and products and between products to obtain a federated learning collaborative filtering recommendation model, and using the federated learning collaborative filtering recommendation model to recommend corresponding products to target users. The present application also provides a product recommendation device, computer equipment and storage medium based on federated learning. In addition, the present application also relates to blockchain technology, and the first user features can be stored in the blockchain. The present application can improve the accuracy of product recommendations while ensuring privacy and security.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a product recommendation method, apparatus, computer equipment and medium based on federated learning. Background Art

[0002] Health insurance has been one of the core driving forces for the development of the insurance industry in recent years, and the big health industry is also developing rapidly. At the same time, the wave of technologies such as big data, cloud computing, artificial intelligence, the Internet of Things, and smart wearables are also impacting the health industry. Therefore, many insurance companies hope to combine these technologies and develop the "big health" industry through health insurance business.

[0003] At present, some insurance companies have launched insurance plans combined with health big data, but there are still many defects and problems. First of all, the data dimensions considered in these plans are relatively single, only considering the user's daily step data or other few health data, and not considering the hospital's more detailed and in-depth data, as well as personal living consumption levels; secondly, they do not have the function of accurate insurance recommendation; finally, the way they collect data may involve privacy issues, and it is difficult to ensure the security of privacy information. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a product recommendation method, device, computer equipment and medium based on federated learning to solve the problem in related technologies that the data collected is of a single dimension and the security of privacy information cannot be guaranteed, and accurate product recommendations cannot be made.

[0005] In order to solve the above technical problems, the embodiment of the present application provides a product recommendation method based on federated learning, which adopts the following technical solution:

[0006] Unite multiple participants to establish a federated learning collaborative filtering recommendation model to be trained;

[0007] Respectively obtain the first user feature corresponding to each participant in the local data and the product feature attribute in the product database;

[0008] Training the to-be-trained federated learning collaborative filtering recommendation model according to the first user feature and the product feature attribute to obtain a federated learning collaborative filtering recommendation model; and

[0009] Use the trained federated learning collaborative filtering recommendation model to recommend corresponding products to target users.

[0010] Furthermore, the step of training the federated learning collaborative filtering recommendation model to be trained according to the first user feature and the product feature attribute includes:

[0011] Calculate the correlation coefficient between the user and the product based on the first user feature;

[0012] Adjusting the weight of the corresponding first user feature according to the correlation coefficient to obtain a second user feature;

[0013] Inputting the second user feature and the product attribute into a federated learning collaborative filtering recommendation model to be trained, calculating a first similarity between products, and clustering the products according to the first similarity;

[0014] Sorting is performed according to the second user feature and the clustering result.

[0015] Furthermore, the step of clustering products according to the first similarity includes:

[0016] Step A: Perform initial clustering on the products in the product database to obtain K initial product classes, which constitute an initial product class set;

[0017] Step B: Calculate the inter-class similarity between every two product classes;

[0018] Step C: Compare the calculated similarities between each category with the preset threshold value;

[0019] Step D: Merge two product classes greater than a preset threshold into a new product class, combine all the merged new product classes and the unmerged initial product classes, and obtain a set of intermediate product classes for secondary clustering;

[0020] Step E: looping through step B to step D until all inter-class similarities in the intermediate product class set are less than or equal to a preset threshold, thereby obtaining a clustered final product class set, wherein the final product class set includes multiple final product classes;

[0021] Step F: Output the final product class set after clustering.

[0022] Furthermore, the step of sorting according to the second user feature and the clustering result includes:

[0023] determining the closest final product category from the clustering results according to the second user feature;

[0024] The recommendations are ranked according to the similarity within the closest final product category.

[0025] Furthermore, the step of determining the product category closest to the user according to the second user feature includes:

[0026] Calculating a second similarity between the second user feature and each type of final product cluster center;

[0027] The final product category corresponding to the cluster center with the greatest similarity is determined as the closest product category.

[0028] Furthermore, the step of using the federated learning collaborative filtering recommendation model to recommend corresponding products to target users includes:

[0029] Obtain the user features of the target user in each participant respectively as the target user features;

[0030] Input the target user characteristics into the federated learning collaborative filtering recommendation model to calculate the recommendation index of the product among each participant;

[0031] Calculate the comprehensive recommendation index of the product according to the recommendation index of each participant;

[0032] Recommend corresponding products to target users according to the comprehensive recommendation index.

[0033] Furthermore, the step of inputting the target user features into the federated learning collaborative filtering recommendation model to calculate the recommendation index of the product among each participant includes:

[0034] Determining a third similarity between the target user and each type of final product according to the target user characteristics;

[0035] The recommendation index of each final product category among each participant is determined according to the ranking of the third similarity.

[0036] In order to solve the above technical problems, the embodiment of the present application also provides a product recommendation device based on federated learning, which adopts the following technical solution:

[0037] Create a module to unite multiple participants to build a federated learning collaborative filtering recommendation model to be trained;

[0038] An acquisition module, used to respectively acquire the first user feature corresponding to the participant's local data and the product feature attribute of the product in the product database;

[0039] a training module, configured to train the to-be-trained federated learning collaborative filtering recommendation model according to the first user feature and the product feature attribute to obtain a trained federated learning collaborative filtering recommendation model; and

[0040] The recommendation module is used to recommend corresponding products to target users using the trained federated learning collaborative filtering recommendation model.

[0041] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0042] The computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the product recommendation method based on federated learning as described above are implemented.

[0043] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0044] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the product recommendation method based on federated learning as described above are implemented.

[0045] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0046] The present application establishes a federated learning collaborative filtering recommendation model to be trained by uniting multiple participants, obtains the first user features corresponding to the local data of all participants and the characteristic attributes of the products in the product database, trains the federated learning collaborative filtering recommendation model to be trained according to the first user features and the characteristic attributes, obtains the trained federated learning collaborative filtering recommendation model, and finally uses the trained federated learning collaborative filtering recommendation model to recommend corresponding products to target users; the present application can realize the diverse dimensions of data collection by establishing a federated learning collaborative filtering recommendation model, and by using the trained federated learning collaborative filtering recommendation model for product recommendations, it can accurately recommend products that better meet customer needs while ensuring privacy and security. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0049] Figure 2 A flowchart of an embodiment of a product recommendation method based on federated learning according to the present application;

[0050] Figure 3 yes Figure 2 A flowchart of a specific implementation of step S203;

[0051] Figure 4 is a structural diagram of an embodiment of a product recommendation device based on federated learning according to the present application;

[0052] Figure 5 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0054] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0055] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0056] In order to solve the problem that the data collected in the related technology is single-dimensional and cannot accurately recommend products while ensuring the security of privacy information, this application provides a product recommendation method based on federated learning, involving artificial intelligence deep learning, which can be applied to Figure 1 In the system architecture 100 shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0057] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0058] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, etc.

[0059] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0060] It should be noted that the product recommendation method based on federated learning provided in the embodiment of the present application is generally executed by a server or a terminal device, and accordingly, the product recommendation device based on federated learning is generally set in a server or a terminal device.

[0061] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0062] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for product recommendation based on federated learning according to the present application. The method for product recommendation based on federated learning comprises the following steps:

[0063] Step S201, uniting multiple participants to establish a federated learning collaborative filtering recommendation model to be trained.

[0064] Federated learning refers to a method of machine learning modeling by uniting different participants (participant, or party, also known as data owner, or client). In federated learning, participants do not need to expose their own data to other participants and coordinators (also known as servers, parameter servers, or aggregation servers). Therefore, federated learning can well protect user privacy and data security, and can solve the problem of data silos.

[0065] In this embodiment, the multiple participants include hospitals, banks, smart wearable devices, and insurance companies.

[0066] Step S202, respectively obtain the first user feature corresponding to each participant in the local data and the product feature attribute in the product database.

[0067] It should be noted that the products include health insurance, accident insurance, car insurance, etc. In this embodiment, health insurance is taken as an example for detailed description.

[0068] In this embodiment, the first user characteristics are obtained from each participant in the federal system. The user characteristics may include real-time physical health conditions, past medical history, the user's physical examination results, consumer purchasing power, and insurance purchase preferences and history; the characteristic attributes of the product may include the user's acceptable annual insurance premium for the insurance product, the preferred insurance type, the preferred insurance company size, insurance period, insurance liability type, etc.

[0069] In this embodiment, the acquired data is the first user feature and the product feature attribute, and the acquired data needs to be preprocessed, specifically including: data cleaning the acquired data, integrating the data of multiple data sources after data cleaning into consistent data by using data integration, and normalizing the integrated consistent data so that each indicator is at the same order of magnitude for subsequent training. Data cleaning is to eliminate incomplete data, erroneous data and duplicate data to achieve the purpose of correcting data.

[0070] It should be noted that when collecting the first user characteristic data, the user ID hash value can be used to collect it, which can protect non-shared users of multiple participants.

[0071] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned first user characteristics, the above-mentioned first user characteristics can also be stored in a node of a blockchain.

[0072] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0073] Step S203: Train the federated learning collaborative filtering recommendation model to be trained according to the first user feature and the product feature attribute to obtain a trained federated learning collaborative filtering recommendation model.

[0074] In a specific implementation of this embodiment, see Figure 3 As shown, the steps of training the federated learning collaborative filtering recommendation model to be trained according to the first user feature and the product feature attribute specifically include:

[0075] Step S301, calculating the correlation coefficient between the user and the product based on the first user feature.

[0076] In this embodiment, if the first user feature is continuous, the Pearson correlation coefficient is used to calculate the correlation coefficient between the user and the product. If the first user feature is a categorical feature, the Kendall concordance coefficient is used to calculate the correlation coefficient between the user and the product.

[0077] Step S302: adjusting the weight of the corresponding first user feature according to the correlation coefficient to obtain the second user feature.

[0078] It should be understood that the products purchased or preferred by the user can be determined based on the first user feature. If the user purchases or prefers a product, its correlation coefficient is large. When the user does not want to see a product that he has purchased, the corresponding user feature weight can be reduced to reduce its correlation coefficient.

[0079] Each first user feature is assigned a different weight. The initial weight of the first user feature can be calculated using the entropy weight method. If the information entropy of a first user feature is smaller, it means that the variation degree of the first user feature is greater, the amount of information provided is more, the role it can play in the comprehensive evaluation is greater, and its weight is greater; on the contrary, the larger the information entropy of a first user feature is, it means that the variation degree of the first user feature is smaller, the amount of information provided is less, the role it plays in the comprehensive evaluation is smaller, and its weight is smaller.

[0080] Step S303: input the second user feature and the product attribute into the federated learning collaborative filtering recommendation model to be trained, calculate the first similarity between the products, and cluster the products according to the first similarity.

[0081] In this embodiment, the first similarity between products is calculated according to product feature attributes, and the algorithm of the first similarity includes but is not limited to the cosine similarity algorithm, the Jaccard coefficient, and the like.

[0082] The method for clustering products according to the first similarity specifically comprises the following steps:

[0083] Step A: Initially cluster the products in the product database to obtain K initial product classes, which constitute an initial product class set.

[0084] The selection of initial cluster centers has a great influence on the clustering results. Once the initial value is not well selected, effective clustering results may not be obtained. Therefore, the method of selecting cluster centers is to ensure that the distance between the initial cluster centers is as far as possible. The cluster centers need to be initialized. The initialization process specifically includes: randomly selecting a product from the product database as the first cluster center, calculating the distance between each product in the product database and the first selected cluster center; calculating the probability of each product being selected as the next cluster center based on the distance, taking a random number r between [0,1], and subtracting the probability of the first data from r. If the subtracted result is less than or equal to 0, it is selected as the next cluster center; conversely, if the result is greater than 0, continue to subtract the probability of the next data, and repeat the process until the subtracted result is less than or equal to 0; repeat the operation until all K initial cluster centers are selected.

[0085] After determining the K initial cluster centers, the Euclidean distance between each product and the initial cluster center will be calculated for the other products in the product database, and they will be divided into the category corresponding to the nearest initial cluster center according to the criterion of the closest distance. The mean value corresponding to all data in each category will be used as the cluster center of the category, and the cluster center will be updated until the cluster center converges and the K final cluster centers are determined. According to the K final cluster centers, the products in the product database will be divided into the category corresponding to the nearest final cluster center according to the criterion of the closest distance. The determined K product categories constitute the set VC = {VC1, VC2, ..., VC k}.

[0086] The basis for judging the convergence of cluster centers is that the distance between the newly calculated cluster center and the cluster center obtained by the previous clustering is less than a preset threshold.

[0087] Step B: Calculate the inter-class similarity between every two product classes.

[0088] In this embodiment, the inter-class similarity is used to measure the similarity between different clustering categories. The larger the value of the inter-class similarity is, the closer the two categories are. On the contrary, the smaller the value of the inter-class similarity is, the greater the difference between the two categories is.

[0089] Step C: Compare the calculated similarities between each category with a preset threshold value.

[0090] It should be understood that the preset threshold can be set according to actual conditions. If the inter-class similarity is greater than the preset threshold, it means that the first product class and the second product class are relatively close, and the two can be aggregated into one category.

[0091] Step D, merge two product classes whose inter-class similarity is greater than a preset threshold into a new product class, combine all the merged new product classes and the unmerged initial product classes, and obtain a set of intermediate product classes for secondary clustering.

[0092] The two product classes are the first product class and the second product class. The calculated inter-class similarity is compared with the preset threshold. If it is greater than the preset threshold, it means that the two product classes are relatively close and can be merged into a new product class. For example, the products are divided into 5 product classes, that is, K is 5, and the product class set is VC = {VC1, VC2, VC3, VC4, VC5}. The inter-class similarity between product classes VC1 and VC2 is greater than the preset threshold. VC2 is added to VC2 to form a new product class VC 12 The intermediate product class set obtained is VC = {VC 12 ,VC3,VC4,VC5}.

[0093] Step E, looping through steps B to D until the inter-class similarity is less than or equal to a preset threshold, thereby obtaining a clustered final product class set, which includes multiple final product classes.

[0094] Step F6, outputting the final product category set after clustering.

[0095] In this embodiment, the final product class set includes multiple final product classes. After the final product class is determined, its cluster center is re-determined, specifically by calculating the average position of all products in each final product class, and determining the average position as the cluster center of the final product class.

[0096] In this embodiment, clustering products according to similarity can improve the accuracy of clustering, thereby ensuring the accuracy of subsequent product recommendations.

[0097] Step 304: sorting according to the second user feature and the clustering result to complete the training of the federated learning collaborative filtering recommendation model.

[0098] Specifically, the final product category closest to the user is determined from the clustering results according to the second user feature, and the final product categories closest to the user are ranked and recommended according to the intra-category similarity of the final product categories closest to the user.

[0099] The method for determining the closest product category according to the second user feature includes: calculating the second similarity between the second user feature and each type of final product category cluster center, and determining the final product category corresponding to the cluster center with the largest similarity as the closest product category.

[0100] In this embodiment, the cosine similarity algorithm can be used to calculate the second similarity. The similarity between two products in the same category has been calculated in step S303, which will not be repeated here. By determining the closest product category based on user characteristics and sorting and recommending according to the similarity within the category, the accuracy of product recommendation is further guaranteed.

[0101] Step S204: Use the trained federated learning collaborative filtering recommendation model to recommend corresponding products to target users.

[0102] After training the federated learning collaborative filtering recommendation model, use the model to accurately recommend corresponding products to target users.

[0103] In some optional implementations of this embodiment, the method of using the federated learning collaborative filtering recommendation model to recommend corresponding products to target users includes the following steps:

[0104] Obtain the user features of the target user in each participant respectively as the target user features;

[0105] Input the target user characteristics into the federated learning collaborative filtering recommendation model to calculate the product recommendation index for each participant;

[0106] Calculate the comprehensive recommendation index of the product according to the recommendation coefficient of each participant;

[0107] Recommend corresponding products to target users based on the comprehensive index.

[0108] The target users are shared users of multiple participants. The federated learning collaborative filtering recommendation model can ensure that the data of each participant is retained locally, ensuring user privacy and security.

[0109] Specifically, the third similarity between the target user and each type of final product is determined according to the target user characteristics, and the recommendation index of each type of final product in each participant is determined according to the ranking of the third similarity.

[0110] It should be noted that the third similarity between the target user and each final product category can be determined by calculating the similarity between the target user feature and the cluster center of each final product category.

[0111] For example, the multiple participants are hospitals, banks, smart wearable devices and insurance companies. Assuming that the final product categories are A, B and C, the target features of the target users in hospitals, banks, smart wearable devices and insurance companies are obtained respectively, and it is calculated that in hospitals, the ranking of product categories is A, B and C from high to low; in banks, the ranking of product categories is A, C and B from high to low; in smart wearable devices, the ranking of smart wearable device product categories is C, B and A from high to low; in insurance companies, the ranking of product categories is B, C and A from high to low, and the score corresponding to each ranking is different. For example, the first place corresponds to a score of 5, and the second place corresponds to a score of 1. The score is 3, and the score corresponding to the third place is 1. The scoring weights of multiple participants are also different. For example, the scoring weight of the hospital is 0.2, the bank is 0.2, the smart wearable device is 0.2, and the insurance company is 0.4. It can be concluded that the comprehensive recommendation index of Class A products is P1=5×0.2+5×0.2+1×0.2+1×0.4=2.6, and the comprehensive recommendation index of Class B products is P2=3×0.2+1×0.2+3×0.2+5×0.4=3.4; the comprehensive recommendation index of Class C products is P1=1×0.2+3×0.2+5×0.2+3×0.4=3.0. Therefore, Class B products are recommended to target users based on the comprehensive recommendation index.

[0112] The present application establishes a federated learning collaborative filtering recommendation model to be trained by uniting multiple participants, obtains the corresponding user features in the local data of all participants and the characteristic attributes of the products in the product database, trains the federated learning collaborative filtering recommendation model to be trained according to the user features and characteristic attributes, obtains the trained federated learning collaborative filtering recommendation model, and finally uses the trained federated learning collaborative filtering recommendation model to recommend corresponding products to target users; the present application can realize the diverse dimensions of data collection by establishing a federated learning collaborative filtering recommendation model, and by using the trained federated learning collaborative filtering recommendation model for product recommendations, it can accurately recommend insurance products that are more in line with the health needs of customers while ensuring privacy and security.

[0113] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0114] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0115] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0116] Further references Figure 4 , as a response to the above Figure 2 The present application provides an embodiment of a product recommendation device based on federated learning. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0117] like Figure 4As shown, the product recommendation device based on federated learning described in this embodiment includes: a creation module 401, an acquisition module 402, a training module 403 and a recommendation module 404. Among them:

[0118] The creation module 401 is used to unite multiple participants to establish a federated learning collaborative filtering recommendation model to be trained;

[0119] The acquisition module 402 is used to respectively acquire the first user feature corresponding to the participant's local data and the product feature attribute of the product in the product database;

[0120] The training module 403 is used to train the federated learning collaborative filtering recommendation model to be trained according to the first user feature and the product feature attribute to obtain a trained federated learning collaborative filtering recommendation model;

[0121] The recommendation module 404 is used to recommend corresponding products to target users using the trained federated learning collaborative filtering recommendation model.

[0122] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned first user characteristics, the above-mentioned first user characteristics can also be stored in a node of a blockchain.

[0123] In this embodiment, the training module 403 includes a calculation submodule, an adjustment submodule, a clustering submodule, and a sorting submodule;

[0124] The calculation submodule is used to calculate the correlation coefficient between the user and the product based on the first user feature;

[0125] The adjustment submodule is used to adjust the weight of the corresponding first user feature according to the correlation coefficient to obtain the second user feature;

[0126] The clustering submodule is used to input the second user feature and the product attribute into the federated learning collaborative filtering recommendation model to be trained, calculate the first similarity between products, and cluster the products according to the first similarity;

[0127] The sorting submodule is used to sort according to the second user characteristics and the clustering results to complete the training of the federated learning collaborative filtering recommendation model.

[0128] In this embodiment, the clustering submodule is further used for:

[0129] Step A: Perform initial clustering on the products in the product database to obtain K initial product classes, which constitute an initial product class set;

[0130] Step B: Calculate the inter-class similarity between every two product classes;

[0131] Step C: Compare the calculated similarities between each category with the preset threshold value;

[0132] Step D: Merge two product classes whose inter-class similarity is greater than a preset threshold into a new product class, and combine all the merged new product classes and the unmerged initial product classes to obtain a set of intermediate product classes for secondary clustering;

[0133] Step E: looping through steps B to D until all inter-class similarities in the intermediate product class set are less than or equal to a preset threshold, thereby obtaining a clustered final product class set, wherein the final product class set includes multiple final product classes;

[0134] Step F: Output the final product class set after clustering.

[0135] In this embodiment, the creation submodule includes a processing unit and a creation unit;

[0136] The processing unit is used to determine the closest final product category from the clustering results according to the second user feature;

[0137] The creation unit is used to perform ranking recommendation according to the intra-class similarity of the closest final product class.

[0138] In this embodiment, the processing unit is further configured to calculate a second similarity between the second user feature and each final product cluster center; and determine the final product class corresponding to the cluster center with the greatest similarity as the closest product class.

[0139] In this embodiment, the recommendation module 404 includes an acquisition submodule, a calculation submodule, and a recommendation submodule;

[0140] The acquisition submodule is used to respectively acquire the user features of the target user in each participant as the target user features;

[0141] The calculation submodule is used to input the target user characteristics into the federated learning collaborative filtering recommendation model to calculate the recommendation index of the product among each participant; and calculate the comprehensive recommendation index of the product according to the recommendation index of each participant;

[0142] The recommendation submodule is used to recommend corresponding products to target users according to the comprehensive recommendation index.

[0143] In this embodiment, the calculation submodule is further used to determine the third similarity between the target user and each type of final product according to the target user characteristics; and determine the recommendation index of each type of final product in each participant according to the ranking of the third similarity.

[0144] The above-mentioned product recommendation device based on federated learning can realize the diverse dimensions of data collection by establishing a federated learning collaborative filtering recommendation model. By using the trained federated learning collaborative filtering recommendation model for product recommendation, it is possible to accurately recommend insurance products that are more in line with the health needs of customers while ensuring privacy and security.

[0145] To solve the above technical problems, the present application also provides a computer device. Figure 5 , Figure 5 This is a basic structural block diagram of the computer device in this embodiment.

[0146] The computer device 5 includes a memory 51, a processor 52, and a network interface 53 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 5 with components 51-53, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0147] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0148] The memory 51 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 51 can be an internal storage unit of the computer device 5, such as a hard disk or memory of the computer device 5. In other embodiments, the memory 51 can also be an external storage device of the computer device 5, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 5. Of course, the memory 51 can also include both the internal storage unit of the computer device 5 and its external storage device. In this embodiment, the memory 51 is generally used to store the operating system and various application software installed on the computer device 5, such as computer-readable instructions of a product recommendation method based on federated learning, etc. In addition, the memory 51 can also be used to temporarily store various types of data that have been output or are to be output.

[0149] The processor 52 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 52 is generally used to control the overall operation of the computer device 5. In this embodiment, the processor 52 is used to run computer-readable instructions stored in the memory 51 or process data, such as computer-readable instructions for running the product recommendation method based on federated learning.

[0150] The network interface 53 may include a wireless network interface or a wired network interface. The network interface 53 is generally used to establish a communication connection between the computer device 5 and other electronic devices.

[0151] This embodiment implements the steps of the product recommendation method based on federated learning in the above embodiment when the processor executes computer-readable instructions stored in the memory, which can realize the diverse dimensions of data collection. At the same time, while ensuring privacy and security, it can accurately recommend insurance products that better meet the health needs of customers.

[0152] The present application also provides another implementation, namely, providing a computer-readable storage medium, which stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor, so that the at least one processor executes the steps of the product recommendation method based on federated learning as described above, which can realize the diverse dimensions of data collection, and at the same time, accurately recommend insurance products that are more in line with the health needs of customers while ensuring privacy and security.

[0153] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0154] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A product recommendation method based on federated learning, characterized in that: The steps include: Unite multiple participants to establish a federated learning collaborative filtering recommendation model to be trained; Respectively obtain the first user feature corresponding to each participant in the local data and the product feature attribute in the product database; Training the to-be-trained federated learning collaborative filtering recommendation model according to the first user feature and the product feature attribute to obtain a trained federated learning collaborative filtering recommendation model; and Use the trained federated learning collaborative filtering recommendation model to recommend corresponding products to target users; The step of training the federated learning collaborative filtering recommendation model to be trained according to the first user feature and the product feature attribute includes: Calculate the correlation coefficient between the user and the product based on the first user feature; Adjusting the weight of the corresponding first user feature according to the correlation coefficient to obtain a second user feature; Inputting the second user feature and the product feature attribute into a federated learning collaborative filtering recommendation model to be trained, calculating a first similarity between products, and clustering the products according to the first similarity; Sorting is performed according to the second user feature and the clustering result, and training of the federated learning collaborative filtering recommendation model is completed; The step of clustering products according to the first similarity comprises: Step A: Perform initial clustering on the products in the product database to obtain K initial product classes, which constitute an initial product class set; Step B: Calculate the inter-class similarity between every two product classes; Step C: Compare the calculated similarities between each category with the preset threshold value; Step D: Merge two product classes whose inter-class similarity is greater than a preset threshold into a new product class, and combine all the merged new product classes and the unmerged initial product classes to obtain a set of intermediate product classes for secondary clustering; Step E: looping through steps B to D until all inter-class similarities in the intermediate product class set are less than or equal to a preset threshold, thereby obtaining a clustered final product class set, wherein the final product class set includes multiple final product classes; Step F: Output the final product class set after clustering; Wherein, before the step of initially clustering the products in the product database to obtain K initial product categories, the step includes: Randomly select a product from the product database as the first cluster center, and calculate the distance between each product in the product database and the selected first cluster center; Calculate the probability of each product being selected as the next cluster center based on the distance. Pick a random number r between [0, 1] and subtract the probability of the first data from r. If the result of the subtraction is less than or equal to 0, it will be selected as the next cluster center. On the contrary, if the result is greater than 0, continue to subtract the probability of the next data, and repeat the process until the result after subtraction is less than or equal to 0; Repeat the operation until all K initial cluster centers are selected.

2. The product recommendation method based on federated learning according to claim 1, characterized in that: The step of sorting according to the second user feature and the clustering result comprises: determining the closest final product category from the clustering results according to the second user feature; The recommendations are ranked according to the similarity within the closest final product category.

3. The product recommendation method based on federated learning according to claim 2, characterized in that: The step of determining the closest final product category from the clustering results according to the second user feature comprises: Calculating a second similarity between the second user feature and each type of final product cluster center; The final product category corresponding to the cluster center with the greatest similarity is determined as the closest product category.

4. The product recommendation method based on federated learning according to claim 3, characterized in that: The step of using the trained federated learning collaborative filtering recommendation model to recommend corresponding products to target users includes: Obtain the user features of the target user in each participant respectively as the target user features; Input the target user characteristics into the federated learning collaborative filtering recommendation model to calculate the recommendation index of the product among each participant; Calculate the comprehensive recommendation index of the product according to the recommendation index of each participant; Recommend corresponding products to target users according to the comprehensive recommendation index.

5. The product recommendation method based on federated learning according to claim 4, characterized in that: The step of inputting the target user features into the federated learning collaborative filtering recommendation model to calculate the recommendation index of the product among each participant comprises: Determining a third similarity between the target user and each type of final product according to the target user characteristics; The recommendation index of each final product category among each participant is determined according to the ranking of the third similarity.

6. A product recommendation device based on federated learning, characterized in that: include: Create a module to unite multiple participants to build a federated learning collaborative filtering recommendation model to be trained; An acquisition module, used to respectively acquire the first user feature corresponding to the participant's local data and the product feature attribute of the product in the product database; A training module, used for training the to-be-trained federated learning collaborative filtering recommendation model according to the first user feature and the product feature attribute to obtain a trained federated learning collaborative filtering recommendation model; and The recommendation module is used to recommend corresponding products to target users using the trained federated learning collaborative filtering recommendation model; Wherein, the training module includes a calculation submodule, an adjustment submodule, a clustering submodule and a sorting submodule; The calculation submodule is used to calculate the correlation coefficient between the user and the product based on the first user feature; The adjustment submodule is used to adjust the weight of the corresponding first user feature according to the correlation coefficient to obtain the second user feature; The clustering submodule is used to input the second user feature and the product feature attribute into the federated learning collaborative filtering recommendation model to be trained, calculate the first similarity between products, and cluster the products according to the first similarity; The sorting submodule is used to sort according to the second user characteristics and the clustering results to complete the training of the federated learning collaborative filtering recommendation model; The clustering submodule is further used for: Step A: Perform initial clustering on the products in the product database to obtain K initial product classes, which constitute an initial product class set; Step B: Calculate the inter-class similarity between every two product classes; Step C: Compare the calculated similarities between each category with the preset threshold value; Step D: Merge two product classes whose inter-class similarity is greater than a preset threshold into a new product class, and combine all the merged new product classes and the unmerged initial product classes to obtain a set of intermediate product classes for secondary clustering; Step E: looping through steps B to D until all inter-class similarities in the intermediate product class set are less than or equal to a preset threshold, thereby obtaining a clustered final product class set, wherein the final product class set includes multiple final product classes; Step F: Output the final product class set after clustering; Wherein, before the step of initially clustering the products in the product database to obtain K initial product categories, the step includes: Randomly select a product from the product database as the first cluster center, and calculate the distance between each product in the product database and the selected first cluster center; Calculate the probability of each product being selected as the next cluster center based on the distance. Pick a random number r between [0, 1] and subtract the probability of the first data from r. If the result of the subtraction is less than or equal to 0, it will be selected as the next cluster center. On the contrary, if the result is greater than 0, continue to subtract the probability of the next data, and repeat the process until the result after subtraction is less than or equal to 0; Repeat the operation until all K initial cluster centers are selected.

7. A computer device comprising a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the product recommendation method based on federated learning as described in any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the product recommendation method based on federated learning as described in any one of claims 1 to 5.

Images