Product recommendation method, device, equipment and storage medium

By recording user access records and extracting feature similarity, combined with collaborative filtering algorithm, the recommendation system solves the problem of insufficient consideration of user factors in insurance product recommendation, improves the targetedness and conversion rate of recommendations, and alleviates the cold start problem.

CN111861760BActive Publication Date: 2024-11-15CHINA PING AN PROPERTY INSURANCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010575425.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-06-22
Publication Date
2024-11-15
Estimated Expiration
2040-06-22

AI Technical Summary

Technical Problem

The insurance product recommendation system lacks consideration of the user's personal comprehensive factors during the recommendation process, resulting in poor recommendation and low conversion rate, especially when the product is cold-started.

Method used

By recording user access records, extracting user's characteristic similarity and historical product similarity, determining candidate target users, and pushing target products based on interest, alleviating the cold start problem of collaborative filtering.

Benefits of technology

It improves the targetedness and conversion rate of product recommendations, effectively alleviates the data sparse problem of collaborative filtering algorithms in cold startup, and enhances user group distinction and recommendation accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111861760B_ABST
    Figure CN111861760B_ABST
Patent Text Reader

Abstract

The present application relates to data analysis, and specifically discloses a product recommendation method, device, equipment and storage medium, the method comprising: upon receiving an access request from a first user, obtaining historical product information corresponding to the first user; based on the historical product information, determining the feature similarity and historical product similarity of multiple different first users, wherein the historical product information is stored in a blockchain; based on the feature similarity and historical product similarity, determining a candidate product information set corresponding to a candidate target user; when there is product information corresponding to an access record corresponding to a candidate target user in the candidate product information set, calculating the interest of the candidate target user in each candidate product, and pushing it according to the interest. When recommending products to users, the cold start problem of ordinary collaborative filtering is effectively alleviated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to a product recommendation method, device, equipment and storage medium. Background Art

[0002] With the development of the Internet and information technology, various business fields are becoming increasingly electronic. For example, in the insurance business, you can purchase insurance online, settle claims online, and receive online recommendations for insurance products.

[0003] At present, commonly used algorithms for insurance product recommendation systems include collaborative filtering, matrix decomposition, logistic regression based on feature engineering, decision trees, etc., all of which use historical data as a reference.

[0004] However, when users search for insurance information online, insurance recommendation platforms usually recommend insurance types that many users have purchased, or use the user's own historical data as a reference. However, there is a lack of consideration of the user's personal comprehensive factors, resulting in poor targeting and low promotion conversion rate. If one type of insurance is used to recommend another target insurance, it involves the problem of product recommendation cold start. Therefore, it is necessary to propose a method to accurately recommend products for current users. Summary of the invention

[0005] The present application provides a product recommendation method, device, equipment and storage medium for data analysis, which adopts a method of recommending one product to another product, and attempts to push another product through the feature similarity obtained from one product, thereby effectively alleviating the cold start problem of ordinary collaborative filtering.

[0006] In a first aspect, the present application provides a product recommendation method, the method comprising:

[0007] Upon receiving an access request from a first user, recording a first access record of the first user to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period;

[0008] Based on the historical product information, determining feature similarities and historical product similarities of a plurality of different first users, wherein the feature similarity is the similarity between feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users;

[0009] Based on the feature similarity and the historical product similarity, determine a candidate target user, and obtain a candidate product information set corresponding to the candidate target user;

[0010] When receiving an access request from the candidate target user, recording a second access record corresponding to the candidate target user;

[0011] When the product information corresponding to the second access record exists in the candidate product information set, the interest of the candidate target user in each candidate product in the candidate product information set is calculated, and the target product is determined according to the interest, and the target product is pushed to the candidate target user.

[0012] In a second aspect, the present application further provides a product recommendation device, the device comprising:

[0013] A first recording module is used to record a first access record of a first user upon receiving an access request from the first user, so as to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period;

[0014] A similarity acquisition module, configured to determine, based on the historical product information, feature similarities and historical product similarities of a plurality of different first users, wherein the feature similarity is the similarity between feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users;

[0015] A target user acquisition module, used to determine candidate target users based on the feature similarity and historical product similarity, and to acquire a candidate product information set corresponding to the candidate target users;

[0016] A second recording module, configured to record a second access record corresponding to the candidate target user upon receiving an access request from the candidate target user;

[0017] The product recommendation module is used to calculate the interest of the candidate target user in each candidate product in the candidate product information set when the product information corresponding to the second access record exists in the candidate product information set, determine the target product based on the interest, and push the target product to the candidate target user.

[0018] In a third aspect, the present application also provides a computer device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the product recommendation method as described above when executing the computer program.

[0019] In a fourth aspect, the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the product recommendation method as described above.

[0020] The present application discloses a product recommendation method, device, equipment and storage medium. When receiving an access request from a first user, the first access record of the first user is recorded, thereby obtaining the historical product information corresponding to the first user; based on the historical product information of the first user, by combining feature extraction and collaborative filtering algorithms, the user's feature parameters are extracted through feature initialization of user data, and feature similarities between different first users are formed, which has better discrimination for user groups. When a candidate target user visits, the second access record corresponding to the candidate target user is recorded, and when product information corresponding to the second access record corresponding to the candidate target user exists in the candidate product information set, the interest of the candidate target user in each candidate product in the candidate product information set is calculated, and the target product is determined according to the interest, and the target product is pushed to the candidate target user. By recommending another product through one product information, the cold start problem of ordinary collaborative filtering is effectively alleviated.

[0021] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. 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.

[0023] Figure 1 This is a schematic diagram of an application scenario of a product recommendation method provided in an embodiment of the present application;

[0024] Figure 2 is a schematic flow chart of a product recommendation method provided in an embodiment of the present application;

[0025] Figure 3 is a flowchart of steps for obtaining feature similarity and historical product similarity of two first users provided by an embodiment of the present application;

[0026] Figure 4 It is a schematic flow chart of the steps of determining a candidate target user and obtaining a candidate product information set corresponding to the candidate target user provided by an embodiment of the present application;

[0027] Figure 5 is a schematic block diagram of a product recommendation device provided in an embodiment of the present application;

[0028] Figure 6A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0030] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0031] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0032] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0033] The embodiments of the present application provide a product recommendation method, device, equipment and storage medium. The product recommendation method can be applied to computer equipment, using one product to recommend another product, and trying to push another product through the feature similarity obtained from one product, which effectively alleviates the cold start problem of ordinary collaborative filtering.

[0034] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0035] See also Figure 1 , Figure 1 Schematic diagram of an application scenario of a product recommendation method provided by an embodiment of the present application. Figure 1 As shown, the application scenario includes at least one terminal.

[0036] Specifically, an application is running on the terminal, which attempts to push another product through the feature similarity obtained from one product, effectively alleviating the cold start problem of ordinary collaborative filtering.

[0037] The terminal may include a smart phone, a tablet computer, a laptop computer or a desktop computer.

[0038] See also Figure 2 , Figure 2 is a schematic flow chart of a product recommendation method provided by an embodiment of the present application. The product recommendation method can be applied to Figure 1 In the computer device in the network, the feature similarity obtained from one product is used to try to push another product, which effectively alleviates the cold start problem of ordinary collaborative filtering.

[0039] like Figure 2 As shown, the product recommendation method specifically includes steps S101 to S105.

[0040] S101. When receiving an access request from a first user, record a first access record of the first user to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period.

[0041] In the prior art, non-auto insurance business of property insurance companies refers to all businesses other than vehicle insurance business, and vehicle insurance business and non-auto insurance business constitute the total business of property insurance companies. With the improvement of people's living standards, many users buy vehicles, so auto insurance is a type of insurance that most people will configure.

[0042] In this case, the insurance business of a property insurance company is taken as an example. The first user is a user who has auto insurance. Correspondingly, the historical product information of the first user includes the first historical product information and the second historical product information. The first historical product information corresponds to auto insurance data, and the second historical product information corresponds to non-auto insurance data. Specifically, non-auto insurance includes critical illness insurance, accident insurance and other insurance types other than auto insurance.

[0043] Specifically, when a user wants to buy an insurance product, he will search for relevant information on the web page to learn about some corresponding product information. At this time, the product recommendation device can record the first access record of the first user, thereby obtaining the historical product information corresponding to the first user.

[0044] In some implementations, recording the first access record of the first user to obtain historical product information corresponding to the first user may include:

[0045] The access record of the first user is recorded in a tracking method; based on the access record, the account information corresponding to the first user is determined; and the consumption information corresponding to the account information is used as the historical product information of the first user.

[0046] Data tracking is a term used in the field of data collection (especially in the field of user behavior data collection), which refers to the technology and implementation process for capturing, processing, and sending specific user behaviors or events, such as the number of times a user clicks on an icon, the length of time a user watches a video, and so on.

[0047] The essence of tracking technology is to first monitor the events during the operation of the software application, and then judge and capture the events that need attention when they occur.

[0048] Therefore, when the first user with the first historical product information browses the insurance product introduction (accesses the specified data) within the preset time period, the user can log in to his or her account, obtain the browsing history and account information of the account, and thus obtain the current user's consumption information, such as the purchase of auto insurance and accident insurance, and use the data of all insurance types purchased by the first user as historical product information.

[0049] S102. Based on the historical product information, determine feature similarities and historical product similarities of multiple different first users, wherein the feature similarity is the similarity between feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users.

[0050] In some embodiments, the historical product information includes first historical product information and second historical product information, wherein the second historical product information is product information in the historical product information of the first user other than the first historical product information; after obtaining the historical product information corresponding to the first user, the feature similarities and historical product similarities of multiple different first users are determined. Preferably, the multiple different first users may be two different first users. After obtaining a first user who has the first historical product information and the second historical product information, another first user who also has the first historical product information and the second historical product information is queried, and the feature similarities and historical product similarities of the two first users are determined.

[0051] It should be noted that after obtaining the first historical product information of a first user, that is, determining that user A is a user with auto insurance, and user B is also a user with auto insurance, then user A and user B are both first users, and the corresponding first historical product information are all auto insurance products.

[0052] Specifically, Figure 3 As shown, Figure 3 The embodiment of the present application provides a flowchart of steps for determining the feature similarities and historical product similarities of multiple different first users. Determining the feature similarities and historical product similarities of multiple different first users may include:

[0053] S102a: construct a portrait model of the first user based on the consumption information.

[0054] Specifically, based on the consumption information of the first user, the first user's auto insurance related insurance records can be queried, including the user information of the policyholder, the insured, and the car owner, the insurance premium and other consumption information, etc. By querying the auto insurance related quotation information, the policyholder's car model data, consumption preferences and other information can be obtained. Through the above data, a certain degree of user portrait model can be constructed.

[0055] S102b. Extract model parameters from the portrait model respectively as a set of feature variables corresponding to the first user.

[0056] Specifically, after determining a first user, another user who also has an auto insurance product is selected as the first user, namely user a and user b. The first historical product information corresponds to the auto insurance data of user a and user b. After obtaining the portrait models corresponding to the two first users respectively, the model parameters of the portrait model can be extracted, and some relatively important features can be selected from the model parameters. For example, data such as the brand of car purchased by the user and the amount of premium are selected to calculate the user's consumption power; data such as the user's gender, age, and insured area are selected to calculate the user's consumption background; data such as the interpersonal relationship between the insured and the insured, the car owner, and the number of seats in the car are selected to calculate factors such as the user's interpersonal relationship and family background.

[0057] Then, we extract a series of corresponding characteristic variables of the user (such as gender, age, car model, car price, insurance premium, spending power, family size, click preference, consumption frequency, etc.) to obtain the characteristic variable set X (X1, X2, X3...X n ), a total of n, set different weight index sets K (K1, K2, K3...K n ).

[0058] S102c: Based on the feature variable set, obtain feature similarities between different first users.

[0059] Specifically, based on the feature variable set, the feature similarities of two different first users, ie, user a and user b, are calculated.

[0060] In some implementations, different weight index sets are set for different feature variables to obtain the feature similarity of two first users. The specific formula 1 is:

[0061]

[0062] in, represents the classification parameters for different feature variables, Q abIt is equal to the sum of the user's different feature classification parameters multiplied by the corresponding weights. The classification parameters of different feature variables, for example, are classified according to the car insurance premium. The car insurance premium of 1,000 is the first level, the car insurance premium of 2,000 is the second level..., thus obtaining the classification parameters of different feature variables φ(X i ).

[0063] S102d. Calculate the collaborative filtering similarity of the second historical product information corresponding to multiple different first users to obtain the historical product similarity corresponding to multiple first users, wherein the first historical product information and the second historical product information can be stored in a blockchain node.

[0064] Specifically, the calculation method of the historical product similarity corresponding to multiple first users is: obtaining the second historical product information corresponding to each different first user; based on the second historical product information, using cosine similarity to calculate the historical product similarity of each different first user.

[0065] In some implementations, taking two first users as an example, second historical product information corresponding to the two first users is obtained; based on the second historical product information, the cosine similarity is used to calculate the historical product similarity of the two first users.

[0066] The second historical product information is the historical product information of the first user (user a and user b) except the first historical product information (auto insurance). The collaborative filtering similarity can use the cosine similarity method to calculate the second product information corresponding to the first user, that is, calculate the purchase similarity of non-auto insurance between user a and user b. Assume that N(a) is the second product information of user a, and N(b) is the second product information of user b.

[0067] Then the similarity of the second product information of two users ab is W ab Formula 2:

[0068]

[0069] Among them, N(a) can be the set of non-auto insurance purchased by user a, N(b) can be the set of non-auto insurance purchased by user b, and the similarity between user a and user b’s non-auto insurance purchase is W ab .

[0070] More specifically, the similarity of the non-auto insurance sets of user a and user b is intersection / union, and the larger the ratio, the higher the similarity of non-auto insurance purchases.

[0071] The auto insurance data, that is, the first historical product information, is the product that the user already has. Through the feature initialization of the existing user data, the user's feature parameters are extracted to form the feature similarity between different users, which has a better distinction between user groups and effectively enhances the targeted push of non-auto insurance. In addition, the feature similarity obtained through auto insurance is used to try to push related non-auto insurance, that is, the push of the second product is recommended through the feature similarity of the first product, which effectively alleviates the cold start problem of ordinary collaboration.

[0072] It should be emphasized that in order to further ensure the privacy and security of the first historical product information and the second historical product information, the first historical product information and the second historical product information can also be stored in the nodes of the blockchain.

[0073] S103: Determine candidate target users based on the feature similarity and historical product similarity, and obtain a candidate product information set corresponding to the candidate target users.

[0074] like Figure 4 As shown, Figure 4 The embodiment of the present application provides a schematic flow chart of the steps of determining a candidate target user based on the feature similarity and the historical product similarity, and obtaining a candidate product information set corresponding to the candidate target user, including:

[0075] S103a: Based on the feature similarity and the historical product similarity, determine a second user set, where the second user set includes a plurality of second users, and use the plurality of second users as candidate target users.

[0076] After obtaining the feature similarity and historical product similarity of user a and user b, find the user closest to user a and user b, rank multiple second users from low to high, and obtain multiple second users whose purchase similarity is close to user a and user b, that is, user b1, user b2, etc. A preset proximity threshold can be set to divide B users of similarity. For example, if the preset proximity threshold is 5, 5 users that are close in sequence from high to low are selected to obtain a set of second users as candidate target users. For example, the purchase similarity of 5 second users is closest to that of user a and user b, and the 5 second users are used as candidate target users.

[0077] In some implementations, when subsequently calculating the interest of the candidate target users in the candidate products, the maximum E (E represents the weighted similarity of user characteristics) may be taken, so that the closest candidate target users are obtained.

[0078] S103b: Acquire the third historical product information corresponding to the candidate target user.

[0079] Specifically, after the candidate target users are determined, a third historical product information set of each candidate target user is obtained.

[0080] The third historical product information corresponds to the second historical product information and is non-auto insurance data. After the non-auto insurance data of the candidate target user is obtained, it is used to recommend non-auto insurance products.

[0081] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned third historical product information, the above-mentioned third historical product information can also be stored in the nodes of the blockchain.

[0082] S103c: Remove the product information corresponding to the second historical product information from the third historical product information to obtain a set of candidate product information corresponding to the candidate target user.

[0083] Specifically, after the third historical product information of the candidate target user is obtained, a candidate product information set corresponding to the candidate target user is obtained based on the third historical product information.

[0084] Furthermore, product information corresponding to the second historical product information is removed from the third historical product information to obtain a set of candidate product information corresponding to the candidate target user.

[0085] For example, the third historical product information is non-auto insurance products, including products 1, 2, 3, and 4, and the second historical product information is non-auto insurance product information, including product 1. Then, the product information corresponding to the second historical product information in the third historical product information is eliminated, and the obtained candidate product information set is products 2, 3, and 4.

[0086] S104: When receiving an access request from the candidate target user, record a second access record corresponding to the candidate target user.

[0087] Specifically, when an access request from a candidate target user is received, the tracking technology can also be used to record the access record corresponding to the candidate target user and obtain the corresponding user account information to determine whether the current user is the candidate target user.

[0088] S105. When the product information corresponding to the second access record exists in the candidate product information set, calculate the interest of the candidate target user in each candidate product in the candidate product information set, determine the target product according to the interest, and push the target product to the candidate target user.

[0089] Specifically, when an access request from the candidate target user is received, the access record corresponding to the candidate target user is recorded. That is, when the candidate target user who has auto insurance enters the corresponding insurance purchase page or advertisement pop-up window, the non-auto insurance that other users with similar characteristics are likely to choose to purchase is calculated based on some existing data of the user. By combining auto insurance data, the data sparsity problem caused by cold start can be more effectively reduced.

[0090] Furthermore, after determining the candidate products, the interest of the candidate target users in each candidate product in the candidate product information set is calculated based on the feature similarity and the historical product similarity; the interest is ranked; if the interest is not less than a preset threshold, the candidate products corresponding to the interest are extracted to obtain a candidate product set, and the candidate product set is determined as the target product; and the target product is recommended to the candidate target user.

[0091] In some implementations, the degree of interest of the candidate target user in each candidate product in the candidate product information set is calculated as Formula 3:

[0092] G(i)=∑ S(p,f) (EQ ab +RW ab )×R bi ;

[0093] Among them, E represents the weighted similarity of features, Q ab represents the feature similarity between user a and user b, R represents the weighted collaborative filtering similarity Wab, R bi It indicates that user b purchases candidate non-auto insurance i, or the evaluation index after the user actually uses it, f is a specific candidate target user in the candidate target user set F, p is a non-auto insurance product in the non-auto insurance product set, and s represents a loop calculation.

[0094] In some implementations, the maximum E may be taken (E represents the weighted similarity of user characteristics), so that several most similar candidate target users are obtained.

[0095] After obtaining the interest ranking of each candidate product corresponding to the candidate target user, a preset threshold can be set. When the interest is greater than or equal to the preset threshold, the candidate product corresponding to the interest is determined as the target product, and the target product is pushed to the candidate target user. The recommendation accuracy is higher. New insurance types can be recommended not only from the purchase collection of similar non-auto insurance users, but also from the purchase collection of users with similar features, providing users with rich choices.

[0096] In some implementations, recommending the target product to the candidate target user includes:

[0097] Obtain product names and keywords corresponding to each candidate product in the target product; list the product names and keywords on the current display page of the candidate target user, so that the candidate target user can click to view detailed product information of the candidate product.

[0098] Specifically, after the candidate products corresponding to the interest level are extracted to obtain a candidate product set, and the candidate product set is determined as the target product, in order to better display the target product to the candidate target user, the product name and keywords corresponding to each candidate product in the target product can be obtained; the product names and keywords are listed on the current display page of the candidate target user, so that the candidate target user can click to view the detailed product information of the candidate product.

[0099] In some embodiments, the information can be displayed on the current browsing page of the candidate target user through a pop-up window or the like. For example, the corresponding products with the top five interest levels can be displayed in a list by listing the product abbreviated names and keywords, or some icons that highlight the product features can be added. When the candidate target user clicks on the product abbreviated name or keyword, the detailed product information of the candidate product will pop up for the candidate target user to read.

[0100] The above embodiment provides a product recommendation method. When receiving an access request from a first user, the first access record of the first user is recorded, thereby obtaining the historical product information corresponding to the first user; based on the historical product information of the first user, by combining feature extraction and collaborative filtering algorithms, the feature parameters of the user are extracted through feature initialization of user data, and feature similarities between different first users are formed, which has better discrimination for user groups. When a candidate target user accesses, the second access record corresponding to the candidate target user is recorded, and when product information corresponding to the second access record corresponding to the candidate target user exists in the candidate product information set, the interest of the candidate target user in each candidate product in the candidate product information set is calculated, and the target product is determined based on the interest, and the target product is pushed to the candidate target user. By recommending another product through one product information, the cold start problem of ordinary collaborative filtering is effectively alleviated.

[0101] like Figure 5 As shown, the product recommendation device 400 includes: a first recording module 401 , a similarity acquisition module 402 , a target user acquisition module 403 , a second recording module 404 , and a product recommendation module 405 .

[0102] The first recording module 401 is used to record a first access record of a first user upon receiving an access request from the first user, so as to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period.

[0103] The similarity acquisition module 402 is used to determine the feature similarity and historical product similarity of multiple different first users based on the historical product information, wherein the feature similarity is the similarity between the feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users.

[0104] The target user acquisition module 403 is used to determine candidate target users based on the feature similarity and historical product similarity, and to acquire a candidate product information set corresponding to the candidate target users.

[0105] The second recording module 404 is configured to record a second access record corresponding to the candidate target user when receiving an access request from the candidate target user.

[0106] The product recommendation module 404 is used to calculate the interest of the candidate target user in each candidate product in the candidate product information set when the product information corresponding to the second access record exists in the candidate product information set, determine the target product based on the interest, and push the target product to the candidate target user.

[0107] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0108] The above-mentioned device can be implemented in the form of a computer program. Figure 6 Runs on the computer device shown.

[0109] See also Figure 6 , Figure 6 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a terminal.

[0110] See also Figure 6 The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0111] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any product recommendation method.

[0112] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0113] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any product recommendation method.

[0114] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0115] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0116] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0117] Upon receiving an access request from a first user, a first access record of the first user is recorded to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period; based on the historical product information, feature similarities and historical product similarities of multiple different first users are determined, wherein the feature similarity is the similarity between feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users; based on the feature similarity and historical product similarity, a candidate target user is determined, and a candidate product information set corresponding to the candidate target user is obtained; upon receiving an access request from the candidate target user, a second access record corresponding to the candidate target user is recorded; when product information corresponding to the second access record exists in the candidate product information set, the interest of the candidate target user in each candidate product in the candidate product information set is calculated, and a target product is determined based on the interest, and the target product is pushed to the candidate target user.

[0118] In some embodiments, the processor implements recording the first access record of the first user to obtain historical product information corresponding to the first user, including:

[0119] The access record of the first user is recorded in a tracking method; based on the access record, the account information corresponding to the first user is determined; and the consumption information corresponding to the account information is used as the historical product information of the first user.

[0120] In some embodiments, the processor implements the historical product information to determine the feature similarities and historical product similarities of a plurality of different first users, including:

[0121] Based on the consumption information, a portrait model of the first user is constructed; model parameters are extracted from the portrait model as a set of feature variables corresponding to the first user; based on the set of feature variables, feature similarities between different first users are obtained; collaborative filtering similarities of second historical product information corresponding to multiple different first users are calculated to obtain similarities of historical products corresponding to multiple different first users, wherein the first historical product information and the second historical product information are stored in a blockchain node.

[0122] In some embodiments, the processor further implements the step of calculating the collaborative filtering similarities of the second historical product information corresponding to the plurality of different first users to obtain the similarities of the historical products corresponding to the plurality of different first users, including:

[0123] The second historical product information corresponding to each different first user is obtained; based on the second historical product information, the historical product similarity of multiple different first users is calculated by using cosine similarity.

[0124] In some embodiments, determining a candidate target user based on the feature similarity and the historical product similarity, and obtaining a candidate product information set corresponding to the candidate target user, includes:

[0125] Based on the feature similarity and the historical product similarity, a second user set is determined, the second user set includes multiple second users, and the multiple second users are used as candidate target users; third historical product information corresponding to the candidate target users is obtained; product information corresponding to the second historical product information in the third historical product information is eliminated, and a candidate product information set corresponding to the candidate target users is obtained.

[0126] In some embodiments, the processor implements the step of calculating the interest of the candidate target user in each candidate product in the candidate product information set, determining a target product according to the interest, and pushing the target product to the candidate target user, including:

[0127] Based on the feature similarity and historical product similarity, the interest of the candidate target user in each candidate product in the candidate product information set is calculated; the interest is ranked; if the interest is not less than a preset threshold, the candidate products corresponding to the interest are extracted to obtain a candidate product set, and the candidate product set is determined as a target product; and the target product is recommended to the candidate target user.

[0128] In some embodiments, the processor implements the step of recommending the target product to the candidate target user, including:

[0129] Obtain product names and keywords corresponding to each candidate product in the target product; list the product names and keywords on the current display page of the candidate target user, so that the candidate target user can click to view detailed product information of the candidate product.

[0130] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any product recommendation method provided in the embodiments of the present application.

[0131] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped on the computer device.

[0132] 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.

[0133] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A product recommendation method, It is characterized in that The method comprises: Upon receiving an access request from a first user, a first access record of the first user is recorded to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period; the access record of the first user is recorded in a burying manner; based on the access record, account information corresponding to the first user is determined; and consumption information corresponding to the account information is used as historical product information of the first user; Based on the historical product information, determine the feature similarity and historical product similarity of multiple different first users, wherein the feature similarity is the similarity between feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users; the historical product information includes first historical product information and second historical product information, wherein the second historical product information is product information other than the first historical product information in the historical product information of the first user; based on the consumption information, build a portrait model of the first user; extract model parameters from the portrait model as the feature variable set corresponding to the first user; based on the feature variable set, obtain the feature similarity between different first users; calculate the collaborative filtering similarity of the second historical product information corresponding to multiple different first users, and obtain the historical product similarity corresponding to multiple different first users, wherein the first historical product information and the second historical product information are stored in the blockchain node; Based on the feature similarity and the historical product similarity, determine a candidate target user, and obtain a candidate product information set corresponding to the candidate target user; When receiving an access request from the candidate target user, recording a second access record corresponding to the candidate target user; When the product information corresponding to the second access record exists in the candidate product information set, the interest of the candidate target user in each candidate product in the candidate product information set is calculated, and the target product is determined according to the interest, and the target product is pushed to the candidate target user; based on the feature similarity and historical product similarity, the interest of the candidate target user in each candidate product in the candidate product information set is calculated; the interest is ranked; if the interest is not less than a preset threshold, the candidate products corresponding to the interest are extracted to obtain a candidate product set, and the candidate product set is determined as the target product; and the target product is recommended to the candidate target user.

2. The method according to claim 1, It is characterized in that The calculating collaborative filtering similarities of the second historical product information corresponding to a plurality of different first users to obtain the historical product similarities corresponding to a plurality of different first users includes: Acquire second historical product information corresponding to different first users respectively; Based on the second historical product information, the cosine similarity is used to calculate the historical product similarities of multiple different first users.

3. The method according to claim 1, It is characterized in that The determining of candidate target users based on the feature similarity and the historical product similarity, and obtaining a candidate product information set corresponding to the candidate target users, includes: Based on the feature similarity and the historical product similarity, determine a second user set, where the second user set includes a plurality of second users, and use the plurality of second users as candidate target users; Acquire third historical product information corresponding to the candidate target user; The product information corresponding to the second historical product information is removed from the third historical product information to obtain a set of candidate product information corresponding to the candidate target user.

4. The method according to claim 1, It is characterized in that The recommending the target product to the candidate target user comprises: Obtain product names and keywords corresponding to each candidate product in the target product; The product names and keywords are listed on the current display page of the candidate target user, so that the candidate target user can click to view the detailed product information of the candidate product.

5. A product recommendation device, It is characterized in that include: A first recording module is configured to, upon receiving an access request from a first user, record a first access record of the first user to obtain historical product information corresponding to the first user, wherein the first user is a user who accesses designated data within a preset time period; record the access record of the first user in a burying manner; determine account information corresponding to the first user based on the access record; and use consumption information corresponding to the account information as historical product information of the first user; a similarity acquisition module, for determining feature similarities and historical product similarities of a plurality of different first users based on the historical product information, wherein the feature similarity is the similarity between feature variable sets of different first users, and the historical product similarity is the collaborative filtering similarity of historical products corresponding to different first users; the historical product information includes first historical product information and second historical product information, wherein the second historical product information is product information other than the first historical product information in the historical product information of the first user; constructing a portrait model of the first user based on the consumption information; extracting model parameters from the portrait model as a feature variable set corresponding to the first user; acquiring feature similarities between different first users based on the feature variable set; calculating collaborative filtering similarities of the second historical product information corresponding to a plurality of different first users, and obtaining historical product similarities corresponding to a plurality of different first users, wherein the first historical product information and the second historical product information are stored in a blockchain node; A target user acquisition module, used to determine candidate target users based on the feature similarity and historical product similarity, and to acquire a candidate product information set corresponding to the candidate target users; A second recording module, configured to record a second access record corresponding to the candidate target user upon receiving an access request from the candidate target user; A product recommendation module is used to calculate the interest of the candidate target user in each candidate product in the candidate product information set when the product information corresponding to the second access record exists in the candidate product information set, determine the target product according to the interest, and push the target product to the candidate target user; calculate the interest of the candidate target user in each candidate product in the candidate product information set based on the feature similarity and historical product similarity; rank the interest; if the interest is not less than a preset threshold, extract the candidate products corresponding to the interest to obtain a candidate product set, and determine the candidate product set as the target product; and recommend the target product to the candidate target user.

6. A computer device, It is characterized in that The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the product recommendation method according to any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the product recommendation method according to any one of claims 1 to 4.