A personalized recommendation method for e-commerce based on big data

Abstract

The embodiment of the present invention provides an e-commerce personalized recommendation method based on big data, which belongs to the field of big data. The method includes obtaining user characteristic information corresponding to the user, and the user characteristic information includes: user basic information, user behavior information, and user context information; according to the user characteristic information, determining a first recommended product from candidate products and displaying the first recommended product to the user; according to the user's operation information based on the first recommended product, determining the confidence level data of the first recommended product; when the confidence level data is greater than a first preset threshold, expanding the first recommended product according to the operation information to obtain a second recommended product; displaying the first recommended product and the second recommended product to the user; the present invention achieves a certain balance between recommending products based on inherent information and exploring new interests of users, greatly improving the accuracy of the e-commerce personalized recommendation method.

Description

[0001] The patent application for this invention is a divisional application. The original application number is 202410641520.4. The application date is May 22, 2024. The name of the invention is an e-commerce personalized recommendation method based on big data. Technical Field

[0002] The present application relates to the field of big data, and in particular to an e-commerce personalized recommendation method based on big data. Background Art

[0003] When users browse e-commerce product pages, the system usually recommends products to them based on their browsing preferences. However, user preferences are dynamic, and existing personalized recommendation methods usually have difficulty in discovering users' new interests, which can easily trap users in information cocoons and lead to homogeneous recommended content. Therefore, in order to strike a balance between recommending products based on inherent information and exploring users' new interests, a more accurate e-commerce personalized recommendation method is urgently needed. Summary of the invention

[0004] The main purpose of the embodiments of the present invention is to provide an e-commerce personalized recommendation method based on big data, aiming to solve the problem in the related art that the personalized recommendation of goods on the e-commerce platform is difficult to discover new interests of users, resulting in single recommendation content and low accuracy.

[0005] The embodiment of the present invention provides an e-commerce personalized recommendation method based on big data, comprising:

[0006] Obtaining user characteristic information corresponding to the user, wherein the user characteristic information includes: user basic information, user behavior information, and user context information;

[0007] Determining a first recommended commodity from candidate commodities according to the user characteristic information and displaying the first recommended commodity to the user;

[0008] Determining confidence level data of the first recommended product according to operation information of the user based on the first recommended product;

[0009] When the confidence level data is greater than a first preset threshold, the first recommended product is expanded according to the operation information to obtain a second recommended product;

[0010] Displaying the first recommended product and the second recommended product to the user;

[0011] When the confidence level data is greater than a first preset threshold, expanding the first recommended product according to the operation information to obtain a second recommended product includes:

[0012] When the confidence level data is greater than a first preset threshold, calculating upper confidence limit data of each target commodity, wherein the target commodity is a candidate commodity other than the first recommended commodity;

[0013] Determine the target product whose upper confidence limit data is greater than a second preset threshold as the second recommended product;

[0014] When the confidence level data is greater than a first preset threshold, calculating the upper confidence limit data of each target commodity includes:

[0015] The upper confidence limit data is calculated according to the following formula:

[0016]

[0017] Among them, UCB i represents the upper confidence limit data of the i-th target product, is the expected profit of the i-th target product, n total is the total number of recommended products, n i is the number of times the i-th target product is recommended, S i Indicates the matching degree between the i-th target product and the user.

[0018] The embodiment of the present invention provides an e-commerce personalized recommendation method based on big data. The present application obtains user characteristic information corresponding to the user, and the user characteristic information includes: user basic information, user behavior information, and user context information. By obtaining the user characteristic information that represents the user's preference as the basis for product recommendation, it is beneficial to the cold start of product recommendation; according to the user characteristic information, a first recommended product is determined from candidate products and the first recommended product is displayed to the user; according to the user characteristic information, the first recommended product displayed to the user is screened, so that the user can browse products that may be of interest, thereby improving the user's browsing experience on the e-commerce platform; according to the user's operation information based on the first recommended product, the first recommended product is determined Confidence level data, determine the size of the confidence interval corresponding to the first recommended product, so as to determine the accuracy of the first recommended product according to the confidence interval; when the confidence level data is greater than a first preset threshold, expand the first recommended product according to the operation information to obtain the second recommended product. If the confidence interval of the first recommended product is large, it means that the range of the first recommended product is too broad and may not accurately reflect the user's preferences. It is necessary to explore the user's new preferences through the second recommended product; display the first recommended product and the second recommended product to the user, so that the user can browse the first recommended product and the second recommended product and perform click, purchase and other operations, so as to obtain feedback on the first recommended product and the second recommended product. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] 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.

[0020] Figure 1 A flowchart of a method for personalized e-commerce recommendation based on big data provided by an embodiment of the present invention;

[0021] Figure 2 A schematic diagram of the module structure of an e-commerce personalized recommendation device based on big data provided by an embodiment of the present invention;

[0022] Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] 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.

[0025] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0026] The embodiment of the present invention provides an e-commerce personalized recommendation method based on big data. The e-commerce personalized recommendation method based on big data can be applied to a terminal device, which can be an electronic device such as a tablet computer, a laptop computer, a desktop computer, a personal digital assistant, and a wearable device. The terminal device can be a server or a server cluster.

[0027] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0028] Please refer to Figure 1 , Figure 1 A flowchart of a method for personalized e-commerce recommendation based on big data provided by an embodiment of the present invention.

[0029] like Figure 1 As shown, the e-commerce personalized recommendation method based on big data includes steps S101 to S105.

[0030] Step S101: Obtain user characteristic information corresponding to a user, wherein the user characteristic information includes: user basic information, user behavior information, and user context information.

[0031] For example, when a new user registers, less data related to the user's preferences can be obtained. At this time, it is necessary to recommend products to the user based on the limited information, i.e., the cold start of product recommendation. Specifically, the recommended products to the user can be determined based on the user's basic information such as age, region, occupation, etc.

[0032] For example, after a user has browsed for a period of time, product recommendations may be made to the user based on the collected user behavior information, such as recommending related products based on products that the user has clicked, collected, or purchased.

[0033] Exemplarily, product recommendations may also be made based on user context information. For example, the purchase history of a user's friends may affect the user's purchase preference. Therefore, with the user's permission, context information related to the user's social network may also be collected.

[0034] Step S102: Determine a first recommended product from candidate products based on the user characteristic information and display the first recommended product to the user.

[0035] Exemplarily, based on the user characteristic information obtained in step S101, a first recommended product to be displayed to the user is determined so that the first recommended product meets the user preference reflected in the user characteristic information, thereby improving the user's browsing experience and purchase intention of the e-commerce platform.

[0036] In some implementations, the user basic information includes: user attribute information, device model information; the user behavior information includes: historical browsing information, historical operation information; the user context information includes: interest tag information, user social information, and determining a first recommended product from candidate products based on the user feature information and displaying the first recommended product to the user includes:

[0037] Performing feature fusion on the user attribute information, device model information, browsing history information, interaction history information, interest tag information, and user social information to obtain a user feature vector;

[0038] According to the user feature vector, the product attribute vector of each candidate product is matched, and the candidate product with a matching degree greater than a preset matching degree is determined as the first recommended product.

[0039] Exemplarily, user features and product features of candidate products are vectorized to obtain user feature vectors and product attribute vectors, and the user feature vectors and product attribute vectors are matched, for example, the cosine distance between the user feature vector and the product attribute vector in space can be calculated, and the matching degree is determined according to the preset distance, and the candidate product with a matching degree greater than a preset matching degree is determined as the first recommended product displayed to the user. The value of the preset matching degree can be determined according to actual needs and is not limited here.

[0040] Exemplarily, user attribute information may include the user's age, gender, region, occupation, etc., device model information may include the type of electronic device used by the user for browsing (e.g., mobile phone, PC, tablet, etc.), brand, model, etc.; historical browsing information may be the length of time the user stays on a product page, historical operation information may be the products on which the user clicks, collects (including adding to a shopping cart, etc.), and purchases; interest tag information may be interest tags selected by the user himself, or interest tags determined based on user characteristics, and user social information may be relevant information about friends added by the user.

[0041] In some implementations, matching the user feature vector with the product attribute vector of each candidate product and determining the candidate product having a matching degree greater than a preset matching degree as the first recommended product includes:

[0042] When the number of the first recommended commodities is less than a preset number, the preset matching degree is reduced based on a preset gradient.

[0043] Exemplarily, the value of the preset matching degree is not fixed. If the number of first recommended products determined based on the preset matching degree is less than the preset number, that is, the number of candidate products with a matching degree greater than the preset matching degree is small, it means that the preset matching degree is set too large. In this case, it is necessary to reduce the value of the preset matching degree so that there can be more first recommended products with a matching degree greater than the preset matching degree among the candidate products, thereby ensuring that there are sufficient first recommended products for users to browse.

[0044] Exemplarily, the size of the preset gradient can be set according to actual needs. When the number of the first recommended products is less than the preset number, the preset gradient is subtracted from the preset matching degree to obtain a reduced preset matching degree.

[0045] Step S103: Determine the confidence level data of the first recommended product according to the user's operation information based on the first recommended product.

[0046] For example, the user's user feature information may change dynamically over time. For example, the user's region may change, and the user's social network and the user's clicks and purchases may also accumulate over time. Therefore, the first recommended product may also change dynamically, and the accuracy may be reduced during the change.

[0047] For example, in order to avoid deviations between the recommended products and the user's preferences, explore the user's new preferences, and avoid the monotony of product recommendations, the embodiment of the present application ensures the accuracy of product recommendations by obtaining confidence level data of the first recommended product in real time.

[0048] In some implementations, determining the confidence level data of the first recommended product according to the operation information of the user based on the first recommended product includes:

[0049] Determining a confidence level interval of the first recommended product according to operation information of the user based on the first recommended product;

[0050] The confidence level interval of the first recommended product is calculated according to the following formula:

[0051] C=CI upper -CI lower ;

[0052]

[0053] Among them, C is the confidence level interval, CI upper is the upper limit of the confidence level interval, CI lower is the lower bound of the confidence level interval, SE is the standard error of the confidence level interval, is the mathematical expectation of the user performing an operation on the first recommended product, α represents the expected confidence level, is a standard normal distribution Percentile, n is the number of first recommended products.

[0054] For example, the confidence level interval is used to quantify the uncertainty of user preferences. It usually gives a range within which the user is expected to purchase items with a certain degree of confidence. If there is a wide confidence level interval for estimating user preferences, it indicates that the understanding of their preferences is not accurate enough, and more types of products need to be recommended to explore the user's preferences.

[0055] Among them, the confidence level interval C passes through the upper limit CI of the confidence level intervalupper and the lower bound of the confidence level interval CI lower The difference between is determined, It is the mathematical expectation of the user's actions on the first recommended product, such as the click rate, collection rate, purchase rate, etc. of the user's click, collection, and purchase actions on the first recommended product during browsing. α represents the expected confidence level, which can be set according to demand. For example, if the credibility (1-α) of the first recommended product is expected to reach 95%, α is set to 5%; The size of can be determined by looking up the standard normal distribution table based on the size of α.

[0056] In some implementations, determining the confidence level data of the first recommended product according to the operation information of the user based on the first recommended product includes:

[0057] Calculate a first product of a mathematical expectation of a click operation performed by the user and a first weight, a second product of a mathematical expectation of a favorite operation performed by the user and a second weight, and a third product of a mathematical expectation of a purchase operation performed by the user and a third weight;

[0058] Calculate the sum of the first product, the second product and the third product to obtain the mathematical expectation of the user performing an operation on the first recommended product

[0059] The third weight is greater than the second weight, and the second weight is greater than the first weight.

[0060] For example, click, purchase and favorite operations reflect different degrees of user preference, so corresponding weights can be set for the mathematical expectations of these three operations. Among them, the purchase operation best reflects the user's preference, so the third weight with the largest value is set for the purchase rate. Similarly, the second weight and first weight corresponding to the favorite rate and click rate decrease in sequence.

[0061] For example, by setting different weights reasonably, the mathematical expectation is improved. accuracy.

[0062] Step S104: when the confidence level data is greater than a first preset threshold, the first recommended product is expanded according to the operation information to obtain a second recommended product.

[0063] Exemplarily, exploring the user's potential new interests through the second recommended product can be regarded as a multi-armed bandit problem. In the e-commerce personalized recommendation method based on big data provided in the embodiment of the present application, the ratio between exploration and exploitation can be balanced. In the embodiment of the present application, "arms" can be regarded as different second recommended products, and the long-term cumulative reward of the second recommended product is maximized through the multi-armed bandit algorithm.

[0064] In some implementations, when the confidence level data is greater than a first preset threshold, expanding the first recommended product according to the operation information to obtain a second recommended product includes:

[0065] When the confidence level data is greater than a first preset threshold, calculating upper confidence limit data of each target commodity, wherein the target commodity is a candidate commodity other than the first recommended commodity;

[0066] The target product whose upper confidence limit data is greater than a second preset threshold is determined as the second recommended product.

[0067] For example, the second recommended product can be determined by the upper confidence bound (UCB) algorithm, where an upper confidence limit data is determined for each target product by the UCB algorithm, and the second recommended product is determined according to the size of the upper confidence limit data. By determining the target product with a higher upper confidence limit data as the second recommended product, exploration of target products that may have high rewards but greater uncertainty is encouraged.

[0068] Exemplarily, in order to avoid duplicate recommendations, the target product is a candidate product other than the first recommended product.

[0069] In some implementations, when the confidence level data is greater than a first preset threshold, calculating the upper confidence limit data of each target commodity includes:

[0070] The upper confidence limit data is calculated according to the following formula:

[0071]

[0072] Among them, UCB i represents the upper confidence limit data of the i-th target product, is the expected profit of the i-th target product, n total is the total number of recommended products, n i is the number of times the i-th target product is recommended, S iIndicates the matching degree between the i-th target product and the user.

[0073] Exemplarily, the upper confidence limit data of the i-th target product is determined based on the operation information of all users who are recommended the product on the e-commerce platform, where is the expected revenue of the i-th target product, which can be determined based on the click rate, collection rate, and purchase rate of users who are recommended the target product; n total It is the total number of recommended products, indicating the cumulative number of all products recommended to users by the e-commerce platform so far.

[0074] For example, in order to make the recommended products relevant to the user's existing interests, the matching degree S between the i-th target product and the user is i As one of the bases for calculating UCB.

[0075] Step S105: display the first recommended product and the second recommended product to the user.

[0076] Exemplarily, the expanded second recommended products are mixed with the first recommended products for display, so as to avoid the singleness of products browsed by users. Users can browse products that they already prefer and products corresponding to potential new interests on one page, thereby improving the diversity of product recommendations and being able to explore users' new interests in a timely manner, forming new interest tags, etc., thereby improving the accuracy of subsequent product recommendations.

[0077] The e-commerce personalized recommendation method based on big data of the embodiment of the present invention obtains user characteristic information corresponding to the user, and the user characteristic information includes: user basic information, user behavior information, and user context information. By obtaining the user characteristic information representing the user's preference as the basis for product recommendation, it is beneficial to the cold start of product recommendation; according to the user characteristic information, a first recommended product is determined from candidate products and the first recommended product is displayed to the user; according to the user characteristic information, the first recommended product displayed to the user is screened, so that the user can browse products that may be of interest to the user, thereby improving the user's browsing experience of the e-commerce platform; according to the user's operation information based on the first recommended product, the confidence level of the first recommended product is determined The confidence interval corresponding to the first recommended product is determined based on the confidence level data, so as to determine the accuracy of the first recommended product according to the confidence interval; when the confidence level data is greater than a first preset threshold, the first recommended product is expanded according to the operation information to obtain a second recommended product. If the confidence interval of the first recommended product is large, it means that the range of the first recommended product is too broad and may not accurately reflect the user's preferences. It is necessary to explore the user's new preferences through the second recommended product; the first recommended product and the second recommended product are displayed to the user, so that the user can browse the first recommended product and the second recommended product and perform operations such as clicking and purchasing, so as to obtain feedback on the first recommended product and the second recommended product.

[0078] See also Figure 2 , Figure 2 An e-commerce personalized recommendation device 200 based on big data is provided in the embodiment of the present application. The e-commerce personalized recommendation device 200 based on big data includes a feature acquisition module 201, a first display module 202, a feedback acquisition module 203, a product expansion module 204, and a second display module 205. Among them, the feature acquisition module is used to obtain user feature information corresponding to the user, and the user feature information includes: user basic information, user behavior information, and user context information; the first display module 202 is used to determine a first recommended product from candidate products according to the user feature information and display the first recommended product to the user; the feedback acquisition module 203 is used to determine the confidence level data of the first recommended product according to the user's operation information based on the first recommended product; the product expansion module 204 is used to expand the first recommended product according to the operation information to obtain a second recommended product when the confidence level data is greater than a first preset threshold; the second display module 205 is used to display the first recommended product and the second recommended product to the user.

[0079] In some implementations, the user basic information includes: user attribute information, device model information; the user behavior information includes: historical browsing information, historical operation information; the user context information includes: interest tag information, user social information, and the first display module 202 determines the first recommended product from the candidate products according to the user characteristic information and displays the first recommended product to the user, for:

[0080] Performing feature fusion on the user attribute information, device model information, browsing history information, interaction history information, interest tag information, and user social information to obtain a user feature vector;

[0081] According to the user feature vector, the product attribute vector of each candidate product is matched, and the candidate product with a matching degree greater than a preset matching degree is determined as the first recommended product.

[0082] In some implementations, when the first display module 202 matches the product attribute vector of each candidate product according to the user feature vector and determines the candidate product with a matching degree greater than a preset matching degree as the first recommended product, it is further configured to:

[0083] When the number of the first recommended commodities is less than a preset number, the preset matching degree is reduced based on a preset gradient.

[0084] In some implementations, when determining the confidence level data of the first recommended product according to the user's operation information based on the first recommended product, the feedback acquisition module 203 is used to:

[0085] Determining a confidence level interval of the first recommended product according to operation information of the user based on the first recommended product;

[0086] The confidence level interval of the first recommended product is calculated according to the following formula:

[0087] C=CI upper -CI lower ;

[0088]

[0089] Among them, C is the confidence level interval, CI upper is the upper limit of the confidence level interval, CI lower is the lower bound of the confidence level interval, SE is the standard error of the confidence level interval, is the mathematical expectation of the user performing an operation on the first recommended product, α represents the expected confidence level, is a standard normal distribution Percentile, n is the number of first recommended products.

[0090] In some implementations, when determining the confidence level data of the first recommended product according to the user's operation information based on the first recommended product, the feedback acquisition module 203 is used to:

[0091] Calculate a first product of a mathematical expectation of a click operation performed by the user and a first weight, a second product of a mathematical expectation of a favorite operation performed by the user and a second weight, and a third product of a mathematical expectation of a purchase operation performed by the user and a third weight;

[0092] Calculate the sum of the first product, the second product and the third product to obtain the mathematical expectation of the user performing an operation on the first recommended product

[0093] The third weight is greater than the second weight, and the second weight is greater than the first weight.

[0094] In some implementations, when the confidence level data is greater than a first preset threshold, the product expansion module 204 expands the first recommended product according to the operation information to obtain a second recommended product, and is used to:

[0095] When the confidence level data is greater than a first preset threshold, calculating upper confidence limit data of each target commodity, wherein the target commodity is a candidate commodity other than the first recommended commodity;

[0096] The target product whose upper confidence limit data is greater than a second preset threshold is determined as the second recommended product.

[0097] In some implementations, when the confidence level data is greater than the first preset threshold, the product expansion module 204 calculates the upper confidence limit number of each target product, for:

[0098] The upper confidence limit data is calculated according to the following formula:

[0099]

[0100] Among them, UCB i represents the upper confidence limit data of the i-th target product, is the expected profit of the i-th target product, n total is the total number of recommended products, n i is the number of times the i-th target product is recommended, S i Indicates the matching degree between the i-th target product and the user.

[0101] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the e-commerce personalized recommendation device 200 based on big data described above can refer to the corresponding process in the aforementioned embodiment of the e-commerce personalized recommendation method based on big data, and will not be repeated here.

[0102] See also Figure 3 , Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present invention.

[0103] like Figure 3 As shown, the terminal device 300 includes a processor 301 and a memory 302, and the processor 301 and the memory 302 are connected via a bus 303, such as an I2C (Inter-integrated Circuit) bus.

[0104] Specifically, the processor 301 is used to provide computing and control capabilities to support the operation of the entire terminal device. The processor 301 can be a central processing unit (CPU), and the processor 301 can 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 gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0105] Specifically, the memory 302 may be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a mobile hard disk.

[0106] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the embodiment of the present invention, and does not constitute a limitation on the terminal device to which the embodiment of the present invention is applied. The specific server may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0107] The processor is used to run a computer program stored in the memory, and implement any one of the big data-based e-commerce personalized recommendation methods provided in the embodiments of the present invention when executing the computer program.

[0108] In one embodiment, the processor is used to run a computer program stored in the memory, and implements the following steps when executing the computer program:

[0109] Obtaining user characteristic information corresponding to the user, wherein the user characteristic information includes: user basic information, user behavior information, and user context information;

[0110] Determining a first recommended commodity from candidate commodities according to the user characteristic information and displaying the first recommended commodity to the user;

[0111] Determining confidence level data of the first recommended product according to operation information of the user based on the first recommended product;

[0112] When the confidence level data is greater than a first preset threshold, the first recommended product is expanded according to the operation information to obtain a second recommended product;

[0113] The first recommended product and the second recommended product are displayed to the user.

[0114] In some implementations, the user basic information includes: user attribute information, device model information; the user behavior information includes: historical browsing information, historical operation information; the user context information includes: interest tag information, user social information, and the processor 301 determines the first recommended product from the candidate products according to the user feature information and displays the first recommended product to the user in the process of:

[0115] Performing feature fusion on the user attribute information, device model information, browsing history information, interaction history information, interest tag information, and user social information to obtain a user feature vector;

[0116] According to the user feature vector, the product attribute vector of each candidate product is matched, and the candidate product with a matching degree greater than a preset matching degree is determined as the first recommended product.

[0117] In some implementations, the processor 301, in the process of matching the user feature vector with the product attribute vector of each candidate product and determining the candidate product having a matching degree greater than a preset matching degree as the first recommended product, is configured to:

[0118] When the number of the first recommended commodities is less than a preset number, the preset matching degree is reduced based on a preset gradient.

[0119] In some implementations, the processor 301, in the process of determining the confidence level data of the first recommended product according to the operation information of the user based on the first recommended product, is configured to:

[0120] Determining a confidence level interval of the first recommended product according to operation information of the user based on the first recommended product;

[0121] The confidence level interval of the first recommended product is calculated according to the following formula:

[0122] C=CI upper -CI lower ;

[0123]

[0124] Among them, C is the confidence level interval, CI upper is the upper limit of the confidence level interval, CI lower is the lower bound of the confidence level interval, SE is the standard error of the confidence level interval, is the mathematical expectation of the user performing an operation on the first recommended product, α represents the expected confidence level, is a standard normal distribution Percentile, n is the number of first recommended products.

[0125] In some implementations, the processor 301, in the process of determining the confidence level data of the first recommended product according to the operation information of the user based on the first recommended product, is configured to:

[0126] Calculate a first product of a mathematical expectation of a click operation performed by the user and a first weight, a second product of a mathematical expectation of a favorite operation performed by the user and a second weight, and a third product of a mathematical expectation of a purchase operation performed by the user and a third weight;

[0127] Calculate the sum of the first product, the second product and the third product to obtain the mathematical expectation of the user performing an operation on the first recommended product

[0128] The third weight is greater than the second weight, and the second weight is greater than the first weight.

[0129] In some implementations, when the confidence level data is greater than a first preset threshold, the processor 301 expands the first recommended product according to the operation information to obtain a second recommended product, and is configured to:

[0130] When the confidence level data is greater than a first preset threshold, calculating upper confidence limit data of each target commodity, wherein the target commodity is a candidate commodity other than the first recommended commodity;

[0131] The target product whose upper confidence limit data is greater than a second preset threshold is determined as the second recommended product.

[0132] In some implementations, when the confidence level data is greater than the first preset threshold, the processor 301 is used to:

[0133] The upper confidence limit data is calculated according to the following formula:

[0134]

[0135] Among them, UCB i represents the upper confidence limit data of the i-th target product, is the expected profit of the i-th target product, n total is the total number of recommended products, n i is the number of times the i-th target product is recommended, S i Indicates the matching degree between the i-th target product and the user.

[0136] It should be noted that technicians in the relevant field can clearly understand that for the convenience and brevity of description, the specific working process of the terminal device described above can refer to the corresponding process in the aforementioned embodiment of the e-commerce personalized recommendation method based on big data, and will not be repeated here.

[0137] An embodiment of the present invention also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of the big data-based e-commerce personalized recommendation methods provided in the description of the embodiment of the present invention.

[0138] The storage medium may be an internal storage unit of the terminal device described in the foregoing embodiment, such as a hard disk or memory of the terminal device. The storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc., equipped on the terminal device.

[0139] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those of ordinary skill in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0140] It should be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0141] The serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. The above are only specific embodiments of the present invention, but the protection scope of the present invention 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 by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A personalized e-commerce recommendation method based on big data, characterized in that: The method comprises: Obtaining user characteristic information corresponding to the user, wherein the user characteristic information includes: user basic information, user behavior information, and user context information; Determining a first recommended commodity from candidate commodities according to the user characteristic information and displaying the first recommended commodity to the user; Determining confidence level data of the first recommended product according to operation information of the user based on the first recommended product; When the confidence level data is greater than a first preset threshold, the first recommended product is expanded according to the operation information to obtain a second recommended product; Displaying the first recommended product and the second recommended product to the user; When the confidence level data is greater than a first preset threshold, expanding the first recommended product according to the operation information to obtain a second recommended product includes: When the confidence level data is greater than a first preset threshold, calculating upper confidence limit data of each target commodity, wherein the target commodity is a candidate commodity other than the first recommended commodity; Determine the target product whose upper confidence limit data is greater than a second preset threshold as the second recommended product; When the confidence level data is greater than a first preset threshold, calculating the upper confidence limit data of each target commodity includes: The upper confidence limit data is calculated according to the following formula: Among them, UCB i represents the upper confidence limit data of the i-th target product, is the expected profit of the i-th target product, n total is the total number of recommended products, n i is the number of times the i-th target product is recommended, S i Indicates the matching degree between the i-th target product and the user.

2. The e-commerce personalized recommendation method based on big data according to claim 1, characterized in that: The user basic information includes: user attribute information, device model information; the user behavior information includes: historical browsing information, historical operation information; the user context information includes: interest tag information, user social information, and determining a first recommended product from candidate products based on the user feature information and displaying the first recommended product to the user includes: Performing feature fusion on the user attribute information, device model information, browsing history information, interaction history information, interest tag information, and user social information to obtain a user feature vector; According to the user feature vector, the product attribute vector of each candidate product is matched, and the candidate product with a matching degree greater than a preset matching degree is determined as the first recommended product.

3. The e-commerce personalized recommendation method based on big data according to claim 2, characterized in that: The matching of the user feature vector with the commodity attribute vector of each candidate commodity and determining the candidate commodity with a matching degree greater than a preset matching degree as the first recommended commodity includes: When the number of the first recommended commodities is less than a preset number, the preset matching degree is reduced based on a preset gradient.

4. The e-commerce personalized recommendation method based on big data according to claim 1, characterized in that: The step of determining the confidence level data of the first recommended product according to the operation information of the user based on the first recommended product includes: Determining a confidence level interval of the first recommended product according to operation information of the user based on the first recommended product; The confidence level interval of the first recommended product is calculated according to the following formula: C=CI upper -CI lower ; Among them, C is the confidence level interval, CI upper is the upper limit of the confidence level interval, CI lower is the lower bound of the confidence level interval, SE is the standard error of the confidence level interval, is the mathematical expectation of the user performing an operation on the first recommended product, α represents the expected confidence level, is a standard normal distribution Percentile, n is the number of first recommended products.

5. The e-commerce personalized recommendation method based on big data according to claim 4, characterized in that: The step of determining the confidence level data of the first recommended product according to the operation information of the user based on the first recommended product includes: Calculate a first product of a mathematical expectation of a click operation performed by the user and a first weight, a second product of a mathematical expectation of a favorite operation performed by the user and a second weight, and a third product of a mathematical expectation of a purchase operation performed by the user and a third weight; Calculate the sum of the first product, the second product and the third product to obtain the mathematical expectation of the user performing an operation on the first recommended product The third weight is greater than the second weight, and the second weight is greater than the first weight.

Images