Product recommendation methods and devices

Abstract

This application discloses a product recommendation method and apparatus. The method includes: collecting attribute and behavioral information of a target user and attribute and behavioral information of all products in a product pool; using a pre-trained product recommendation model to determine target recommended products matching the target user; the model training samples include attribute and behavioral information of multiple users and attribute and behavioral information of multiple products; analyzing the data using a collaborative filtering algorithm to calculate a first similarity between users and a second similarity between products; determining each user's interest rating for each product, user preferences, and the recommendation priority of each product for each user based on the first and second similarities, and determining a corresponding recommended product for each user; and recommending the target recommended product to the target user based on the target user's attribute information using a selected target product recommendation method. This application solves the technical problems of related product recommendation schemes failing to meet users' personalized needs and exhibiting poor recommendation results.

Description

Technical Field

[0001] This application relates to the field of big data processing technology, and more specifically, to a product recommendation method and apparatus. Background Technology

[0002] With the rapid development of the telecommunications industry and the deepening of digital transformation, telecom operators face the challenge of effectively managing a vast pool of products and accurately providing users with recommendations for telecommunications goods. Traditional recommendation systems often rely on relatively simple user behavior analysis, such as browsing history and purchase records, employing content filtering or collaborative filtering algorithms based on users or items. However, such recommendation methods suffer from the following problems: traditional recommendation algorithms often fail to fully consider users' personalized needs and preferences, meaning that recommended products may not match users' actual needs and interests, resulting in poor recommendation performance and low user satisfaction. Users' needs and preferences may change over time and due to various factors such as environment, and traditional recommendation algorithms often struggle to capture these uncertainties in a timely manner, leading to inflexible and inaccurate recommendation results. Existing recommendation systems often struggle to explain the rationale behind recommendations to users, resulting in a lack of understanding and trust in the recommendation results, which limits the application scope and user acceptance of recommendation systems.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides a product recommendation method and apparatus to at least solve the technical problems that related product recommendation schemes are unable to meet users' personalized needs and have poor recommendation effects.

[0005] According to one aspect of the embodiments of this application, a product recommendation method is provided, comprising: acquiring target user attribute information and target user behavior information of a target user, and target product attribute information and target product behavior information of all products in a product pool; analyzing the target user attribute information, target user behavior information, target product attribute information, and target product behavior information using a pre-trained product recommendation model to obtain target recommended products corresponding to the target user, wherein the training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products, and the sample labels for training the product recommendation model are recommended products corresponding to each user determined by the following method: analyzing user attribute information, user behavior information, product attribute information, and product behavior information using a collaborative filtering algorithm to obtain a first similarity among multiple users and a second similarity among multiple products, and determining each user's interest rating, user preference, and recommendation priority of each product for each user based on the first similarity and the second similarity, and determining the recommended product corresponding to each user based on the interest rating, user preference, and recommendation priority; determining the target product recommendation method corresponding to the target user attribute information, and recommending the target recommended product to the target user based on the target product recommendation method.

[0006] Optionally, the training process of the product recommendation model includes: constructing an initial recommendation model based on deep learning; acquiring user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products, and using the user attribute information and user behavior information of each user, and the product attribute information and product behavior information of multiple products as a set of training samples, and determining the sample label corresponding to each set of training samples; iteratively training the initial recommendation model using multiple sets of training samples and sample labels, and optimizing the model parameters using the stochastic gradient descent method; and using the trained model as the product recommendation model when the preset training termination condition is met.

[0007] Optionally, the sample labels corresponding to each group of training samples are determined, including: constructing a user-product co-occurrence matrix based on user behavior information of multiple users and product behavior information of multiple products, wherein the elements in the user-product co-occurrence matrix are the comprehensive ratings of each user for each product, and the comprehensive rating is a weighted score calculated by weighting the number of different behaviors of each user for each product and the evaluation scores; based on the user-product co-occurrence matrix, using a user-based collaborative filtering algorithm to determine the first similarity between every two users and the first interest rating of each user for each product; using a product-based collaborative filtering algorithm to determine the second similarity between every two products and the second interest rating of each user for each product; for each user and each product, determining the user's preference coefficient for the product based on the first user rating and the first user similarity, wherein the first user rating is a user-product... The product co-occurrence matrix is ​​used to calculate the overall user ratings for products. The first user similarity is the average of the first similarities between the user and all other users who have interacted with the product. The product recommendation priority coefficient is determined based on the first user rating, first product rating, first user similarity, and first product similarity. The first product rating is the average of all user ratings for the product in the user-product co-occurrence matrix, and the first product similarity is the average of the second similarities between the product and other products interacting with the user. The product recommendation rating is determined based on the preference coefficient, recommendation priority coefficient, first interest rating, and second interest rating. For each user, multiple products are ranked from highest to lowest based on their recommendation ratings. A predetermined number of products at the top of the ranking are designated as recommended products for the user, and these recommended products are used as the sample labels for the user's training samples.

[0008] Optionally, the formula for calculating the overall score is: r ui =ω1·b1+ω2·b2+ω3·b3+ω4·b4+ω5·s4, where, r ui This represents user u's overall rating of product i, where ω1, ω2, ω3, ω4, and ω5 are preset weight coefficients, and b1, b2, b3, b4, and s4 represent user u's number of views, purchases, searches, ratings, and ratings for product i, respectively. The formula for calculating the first similarity is: In the formula, sim(u,v) represents the first similarity between user u and user v, A represents the set of items for which both user u and user v have performed actions, and r uk This represents user u's overall rating of product k, r vk This represents user v's overall rating of product k; the formula for calculating the first interest rating is: In the formula, P ui1Let sim(u,n) represent the first interest rating of user u for product i, N(u) represent the set of N users with the highest first similarity to user u, and r represent the first similarity between user u and user n. ni This represents the overall rating of user n for product i; the formula for calculating the second similarity is: In the formula, sim(i,j) represents the second similarity between product i and product j, B represents the set of users who have interacted with both product i and product j, and r ni R represents the overall rating of user n for product i. nj This represents user n's overall rating for product j; the formula for calculating the second interest rating is: In the formula, P ui2 Let sim(i,k) represent the second interest rating of user u for product i, N(i) represent the set of N products with the highest second similarity to product i, sim(i,k) represent the second similarity between product i and product k, and r uk This represents user u's overall rating of product k; the formula for calculating the preference coefficient is: In the formula, α ui r represents the preference coefficient of user u for product i. ui Let represent the overall rating of user u for product i, C represent the set of other users besides user u who have interacted with product i, sim(u,n) represent the first similarity between user u and user n, and c1 and c2 are preset constants; the formula for calculating the recommendation priority coefficient is: In the formula, β ui Let represent the recommendation priority coefficient of product i for user u, C represent the set of other users besides user u who have interacted with product i, sim(u,n) represent the first similarity between user u and user n, D represent the set of other products besides product i that user u has interacted with, sim(i,k) represent the second similarity between product i and product k, and r ui Let r represent the overall rating of user u for product i, E represent the set of users who have interacted with product i, and r represent the overall rating of user u for product i. ni Let S represent the overall rating of user n for product i, where c3 and c4 are preset constants; the formula for calculating the recommendation rating is: S ui =α ui ·β ui ·θ1·P ui1 ·θ2·P ui2 In the formula, S ui α represents the recommendation rating of product i for user u. ui β represents the preference coefficient of user u for product i. ui P represents the recommendation priority coefficient of product i for user u. ui1 P represents user u's first interest rating for product i.ui2 θ1 and θ2 represent the second interest rating of user u for product i, and are preset weight coefficients.

[0009] Optionally, the target user attribute information includes: the target user's identity identifier, age, and gender. Determining the target product recommendation method corresponding to the target user attribute information includes: querying the product recommendation method list for the product recommendation method corresponding to the target user's identity identifier, wherein the product recommendation method list stores product recommendation methods corresponding to different users' identity identifiers; if a product recommendation method corresponding to the target user's identity identifier exists, the product recommendation method is used as the target product recommendation method; if no product recommendation method corresponding to the target user's identity identifier exists, a template recommendation method matching the target user's age and gender is determined from the product recommendation method list and used as the target product recommendation method; in response to the target user's adjustment of the template recommendation method, the adjusted template recommendation method is stored as the product recommendation method corresponding to the target user's identity identifier in the product recommendation method list.

[0010] Optionally, after recommending target products to target users based on the target product recommendation method, the method further includes: obtaining the click-through rate and purchase conversion rate of target users for target recommended products, and determining the ratio of purchase conversion rate to click-through rate; using target user attribute information, target user behavior information, target product attribute information, and target product behavior information as new training samples, and in the process of determining the sample labels of the new training samples, using the ratio as an additional adjustment coefficient to determine the recommendation rating of the target recommended product for the target user; periodically using the new training samples and corresponding sample labels to iteratively train the product recommendation model and update the model parameters.

[0011] Optionally, user attribute information includes at least one of the following: user's age, gender, address, and transactions completed; user behavior information includes at least one of the following: user's browsing history, purchase history, search history, and rating history; product attribute information includes at least one of the following: product type, price, brand, and network standard; and product behavior information includes at least one of the following: product browsing history, purchase history, search history, and rating history.

[0012] According to another aspect of the embodiments of this application, a product recommendation device is also provided, comprising: an acquisition module, configured to acquire target user attribute information and target user behavior information of a target user, and target product attribute information and target product behavior information of all products in a product pool; and an analysis module, configured to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information using a pre-trained product recommendation model to obtain target recommended products corresponding to the target user, wherein the training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products, for training the product recommendation model. The sample labels are the recommended products corresponding to each user, determined in the following way: A collaborative filtering algorithm is used to analyze user attribute information, user behavior information, product attribute information, and product behavior information to obtain the first similarity among multiple users and the second similarity among multiple products. Based on the first and second similarities, each user's interest rating for each product, user preference, and recommendation priority for each product to each user are determined. The recommended products corresponding to each user are then determined based on the interest rating, user preference, and recommendation priority. The recommendation module is used to determine the target product recommendation method corresponding to the target user's attribute information and recommend target recommended products to the target user based on the target product recommendation method.

[0013] According to another aspect of the embodiments of this application, a computer program product is also provided, the computer program product comprising: a computer program, wherein the computer program, when executed by a processor, implements the above-described product recommendation method.

[0014] According to another aspect of the embodiments of this application, an electronic device is also provided, the electronic device including: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-described product recommendation method through the computer program.

[0015] In this embodiment, by collecting target user attribute information (including age, gender, identity, etc.) and behavioral information (such as purchase and browsing history), combined with detailed attributes and market behavior data of products in the product pool, a pre-trained product recommendation model can generate highly personalized product recommendations. During the training process of the pre-trained product recommendation model, a collaborative filtering algorithm is used to calculate the first similarity between users and the second similarity between products. This not only helps the model identify common preferences within user groups but also understands the correlation between products. By combining similarity with user product interest ratings, user preferences, and product recommendation priorities, the model can generate more optimized recommendation decisions, ensuring that recommended products... Not only does it meet users' personalized needs, but it also possesses high market appeal and potential value. The model design incorporates a dynamic feedback mechanism that learns and adjusts from actual user feedback on recommended products (such as click-through rates and purchase conversion rates). By periodically utilizing this feedback information to iterate model training, it continuously updates model parameters, improving prediction accuracy and recommendation effectiveness. The model not only optimizes product recommendations but also supports the determination of personalized recommendation methods. It can intelligently match the most suitable recommendation channels based on user identity, age, gender, and other information, making the recommendation service more closely aligned with user habits and preferences. This solves the technical problem that related product recommendation schemes struggle to meet users' personalized needs and achieve poor recommendation results. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0017] Figure 1 This is a flowchart illustrating an optional product recommendation method according to an embodiment of this application;

[0018] Figure 2 This is a schematic diagram of an optional product recommendation device according to an embodiment of this application;

[0019] Figure 3 This is a schematic diagram of the structure of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0021] It should be noted that the terms "first," "second," etc., used in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] To better understand the embodiments of this application, the following is a translation and explanation of some nouns or terms that appear in the description of the embodiments of this application:

[0023] Collaborative filtering (CF) is a widely used technique in recommender systems that aims to provide personalized recommendations by analyzing users' historical behavior and preferences. The core idea of ​​collaborative filtering is that "similar users like similar items," or "similar items are liked by similar users." Depending on the implementation, collaborative filtering can be divided into user-based collaborative filtering (CF), item-based collaborative filtering (CF), and more complex variations such as matrix factorization.

[0024] Example 1

[0025] According to an embodiment of this application, a product recommendation method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0026] Figure 1 This is a flowchart illustrating a product recommendation method provided in an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0027] Step S102: Obtain the target user attribute information and target user behavior information of the target user, and the target product attribute information and target product behavior information of all products in the product pool;

[0028] Step S104: Analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information using a pre-trained product recommendation model to obtain the target recommended products corresponding to the target users. The training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined in the following way: Analyze the user attribute information, user behavior information, product attribute information, and product behavior information using a collaborative filtering algorithm to obtain the first similarity between multiple users and the second similarity between multiple products. Based on the first and second similarities, determine each user's interest rating for each product, user preference, and recommendation priority for each product to each user. Based on the interest rating, user preference, and recommendation priority, determine the recommended products corresponding to each user.

[0029] Step S106: Determine the target product recommendation method corresponding to the target user attribute information, and recommend target recommended products to the target user based on the target product recommendation method.

[0030] The following section explains each step of the product recommendation method in conjunction with the specific implementation process.

[0031] First, we acquire the target user's attribute information and target user behavior information, as well as the target product's attribute information and target product behavior information for all products in the product pool. By gaining a deeper understanding of the characteristics of the target user and all products in the product pool, we can generate more accurate product recommendations.

[0032] As an optional implementation, user attribute information includes at least one of the following: user's age, gender, address, and services already processed; user behavior information includes at least one of the following: user's browsing history, purchase history, search history, and rating history; product attribute information includes at least one of the following: product type, price, brand, and network standard; product behavior information includes at least one of the following: product browsing history, purchase history, search history, and rating history.

[0033] After obtaining the above information, a pre-trained product recommendation model is used to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information to obtain the target recommended products corresponding to the target users. The training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined in the following way: the collaborative filtering algorithm is used to analyze the user attribute information, user behavior information, product attribute information, and product behavior information to obtain the first similarity among multiple users and the second similarity among multiple products. Based on the first and second similarities, the interest rating, user preference, and recommendation priority of each product for each user are determined. Based on the interest rating, user preference, and recommendation priority, the recommended products corresponding to each user are determined.

[0034] Generally, user-based collaborative filtering methods involve the following steps: calculating the similarity between the target user and other users, commonly using methods such as cosine similarity and Pearson correlation coefficient; predicting the current user's rating for unrated items based on ratings from similar users, typically using a weighted average method with weights equal to the similarity between users; and recommending the highest-rated item to the user based on the predicted rating. Product-based collaborative filtering algorithms also involve the following steps: calculating the similarity between items, commonly using methods such as cosine similarity and Pearson correlation coefficient; predicting the user's rating for unrated items based on ratings from similar items, again using a weighted average method with weights equal to the similarity between items; and recommending the highest-rated item to the user based on the predicted rating. This embodiment utilizes the corresponding collaborative filtering algorithm when calculating the first and second similarities.

[0035] As an optional implementation method, the training process of the product recommendation model can be carried out in the following steps:

[0036] S1, Construct an initial recommendation model based on deep learning;

[0037] S2, obtain user attribute information and user behavior information of multiple users, product attribute information and product behavior information of multiple products, and use the user attribute information and user behavior information of each user, and the product attribute information and product behavior information of multiple products as a set of training samples, and determine the sample label corresponding to each set of training samples.

[0038] S3 uses multiple sets of training samples and sample labels to iteratively train the initial recommendation model, and uses the stochastic gradient descent method to optimize the model parameters. When the preset training termination conditions are met, such as the number of iterations reaching a threshold or the gradient no longer decreasing, the trained model is used as the product recommendation model.

[0039] As an optional implementation, during model training, the following process can be used to determine the sample labels corresponding to each group of training samples:

[0040] S1. Construct a user-product co-occurrence matrix based on user behavior information of multiple users and product behavior information of multiple products. The elements in the user-product co-occurrence matrix are the comprehensive ratings of each user for each product. The comprehensive rating is the score calculated by weighting the number of different behaviors of each user for each product and the rating score. Here, the different behaviors of users for each product refer to behaviors such as browsing, purchasing, searching or rating each product.

[0041] The user-product co-occurrence matrix is ​​a data structure used to reflect the strength of the association between users and products. Rows in the matrix represent users, columns represent products, and element values ​​represent the user's overall rating of the product. The overall rating is obtained by weighting the user's different behaviors towards the product (such as purchase frequency, browsing frequency, etc.) and rating scores (such as a five-star rating). For example, a user might browse product A multiple times but not purchase it, giving it a high rating; purchase product B once, giving it a low rating; and have no interaction with product C. This information is integrated into the user-product co-occurrence matrix for subsequent analysis. As an optional implementation method, the formula for calculating the overall rating is:

[0042] r ui =ω1·b1+ω2·b2+ω3·b3+ω4·b4+ω5·s4

[0043] In the formula, r ui This represents the overall rating of user u for product i. ω1, ω2, ω3, ω4, and ω5 are preset weighting coefficients, and b1, b2, b3, b4, and s4 represent the number of times user u viewed, purchased, searched, rated, and rated product i, respectively.

[0044] S2, based on the user-item co-occurrence matrix, a user-based collaborative filtering algorithm is used to determine the first similarity between every two users and the first interest rating of each user for each item. The similarity is based on the similarity of user behavior; for example, if two users frequently purchase the same product, their similarity is high. The interest rating is based on the user's direct rating of the item and the frequency of the behavior. In this embodiment, the formula for calculating the first similarity is:

[0045]

[0046] In the formula, sim(u,v) represents the first similarity between user u and user v, A represents the set of items for which both user u and user v have performed actions, and r ukThis represents user u's overall rating of product k, r vk This represents user v's overall rating of product k.

[0047] The formula for calculating the first interest rating is:

[0048]

[0049] In the formula, P ui1 Let sim(u,n) represent the first interest rating of user u for product i, N(u) represent the set of N users with the highest first similarity to user u, and r represent the first similarity between user u and user n. ni This represents the overall rating of user n for product i.

[0050] S3 uses a product-based collaborative filtering algorithm to determine the second similarity between every two products and the second interest rating of each user for each product.

[0051] The formula for calculating the second similarity is as follows:

[0052]

[0053] In the formula, sim(i,j) represents the second similarity between product i and product j, B represents the set of users who have interacted with both product i and product j, and r ni R represents the overall rating of user n for product i. nj This represents user n's overall rating of product j.

[0054] The formula for calculating the second interest rating is:

[0055]

[0056] In the formula, P ui2 Let sim(i,k) represent the second interest rating of user u for product i, N(i) represent the set of N products with the highest second similarity to product i, sim(i,k) represent the second similarity between product i and product k, and r uk This represents user u's overall rating of product k.

[0057] S4, for each user and each product, the user's preference coefficient for the product is determined based on the first user rating and the first user similarity. The first user rating is the comprehensive rating of the user for the product in the user-product co-occurrence matrix, and the first user similarity is the average of the first similarities between the user and all other users who have interacted with the product. The formula for calculating the preference coefficient in this embodiment is as follows:

[0058]

[0059] In the formula, α ui r represents the preference coefficient of user u for product i. ui Let represent the overall rating of user u for product i, let C represent the set of users other than user u who have interacted with product i, let sim(u,n) represent the first similarity between user u and user n, and let c1 and c2 be preset constants.

[0060] Among them, molecule r ui This represents user u's overall rating of product i. This rating directly reflects the user's preference for the product; a higher rating indicates greater user interest in the product. C1 is added to the denominator primarily to prevent the denominator from being zero (i.e., when user u is completely dissimilar to other users, to avoid dividing by zero). Furthermore, C1 can adjust the sensitivity of the formula, ensuring that even with low similarity, the preference coefficient is not excessively high. Overall... The meaning is to divide the user's overall rating by the average similarity between user u and other users who have interacted with product i. If user u is very similar to other users who have interacted with product i (i.e., the average similarity is high), then user u's rating may be influenced by these similar users, so the preference coefficient will be relatively small. Conversely, if user u is not similar to other users, then user u's rating is more representative of personal preference, and the preference coefficient will be larger. The existence of c2 ensures that the preference coefficient will not fall below the preset lower limit. This means that even if the user's rating is low or the similarity with other users is high, the preference coefficient will not be too small, thus ensuring a certain minimum preference value. This helps to avoid unreasonable recommendations in extreme cases.

[0061] S5. Determine the product recommendation priority coefficient for the user based on the first user rating, the first product rating, the first user similarity, and the first product similarity. The first product rating is the average of all user ratings for the product in the user-product co-occurrence matrix. The first product similarity is the average of the second similarities between the product and other products that generated the user's behavior. The formula for calculating the recommendation priority coefficient is:

[0062]

[0063] In the formula, β ui Let represent the recommendation priority coefficient of product i for user u, C represent the set of other users besides user u who have interacted with product i, sim(u,n) represent the first similarity between user u and user n, D represent the set of other products besides product i that user u has interacted with, sim(i,k) represent the second similarity between product i and product k, and r ui Let r represent the overall rating of user u for product i, E represent the set of users who have interacted with product i, and r represent the overall rating of user u for product i. niThis represents the overall rating of user n for product i, where c3 and c4 are preset constants.

[0064] It should be noted that, This represents the average similarity between user u and other users who have interacted with product i. It reflects the overall similarity between user u and other users who have interacted with product i. This represents the average similarity between product i and other products that user u has interacted with. It reflects the similarity between product i and other products that user u has previously encountered, and helps to capture users' interest in similar products. This represents the average rating of all users who have interacted with product i. It reflects the overall popularity of product i, i.e., the average evaluation of product i by all users. This represents the product of user similarity and product similarity. This product term emphasizes the combined effect of dual similarity between users and products. That is, if user u is very similar to other users who have interacted with product i, and product i is also very similar to other products that user u has previously encountered, then product i will have a higher recommendation priority for user u. This represents the product of the user's rating of product i and the average rating of product i. This product term takes into account the user's personal preference for product i and the overall popularity of product i. If the user's rating of product i is high and product i is also generally well received by other users, then the product's recommendation priority will be higher. c3 is mainly to prevent the numerator from being zero and can adjust the sensitivity of the recommendation priority coefficient.

[0065] S6. Based on the preference coefficient, recommendation priority coefficient, first interest score, and second interest score, determine the product's recommendation score for the user. The formula for calculating the recommendation score is:

[0066] S ui =α ui ·β ui ·θ1·P ui1 ·θ2·P ui2

[0067] In the formula, S ui α represents the recommendation rating of product i for user u. ui β represents the preference coefficient of user u for product i. ui P represents the recommendation priority coefficient of product i for user u. ui1 P represents user u's first interest rating for product i. ui2 θ1 and θ2 represent the second interest rating of user u for product i, and are preset weight coefficients.

[0068] S7. For each user, sort multiple products from highest to lowest based on the user's recommendation ratings for multiple products, determine the top-ranked preset number of products as the user's recommended products, and use the recommended products as the sample labels for the user's corresponding training samples.

[0069] After training the product recommendation model using the above method, the model is used to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information to obtain the target recommended products corresponding to the target user.

[0070] After obtaining the target recommended products corresponding to the target users, determine the target product recommendation method corresponding to the target user's attribute information, and recommend the target recommended products to the target users based on the target product recommendation method.

[0071] As an optional implementation, the target user attribute information includes: the target user's identity, age, and gender. The process of determining the target product recommendation method corresponding to the target user's attribute information can take the following steps: querying the product recommendation method list for the product recommendation method corresponding to the target user's identity, wherein the product recommendation method list stores product recommendation methods corresponding to different user identity; if a product recommendation method corresponding to the target user's identity exists, the product recommendation method is used as the target product recommendation method; if no product recommendation method corresponding to the target user's identity exists, a template recommendation method matching the target user's age and gender is determined from the product recommendation method list as the target product recommendation method. The template recommendation method is a recommendation strategy pre-set based on the common preferences of the user group. For example, younger users may prefer to receive recommendations through mobile applications, while older users may prefer email or SMS; in response to the target user's adjustment of the template recommendation method, the adjusted template recommendation method is stored as the product recommendation method corresponding to the target user's identity in the product recommendation method list.

[0072] As an optional implementation, after recommending target products to target users based on the target product recommendation method, the following steps can be taken to further collect feedback data from target users on the recommended products in order to optimize the performance of the recommendation model:

[0073] Obtain the click-through rate and purchase conversion rate of target users for target recommended products, and determine the ratio of purchase conversion rate to click-through rate; use target user attribute information, target user behavior information, target product attribute information, and target product behavior information as new training samples, and in the process of determining the sample labels of the new training samples, use the ratio as an additional adjustment coefficient to determine the recommendation rating of the target recommended product for the target user; periodically use the new training samples and corresponding sample labels to iteratively train the product recommendation model and update the model parameters.

[0074] Click-through rate (CTR) refers to the ratio of the number of times a user clicks on a product in the recommendation list to the total number of times that product is displayed to the user. In the context of telecom product recommendations, CTR reflects the user's level of interest in the recommended products and is a quantitative indicator of the initial response to the recommendation results. Purchase conversion rate (PCR) refers to the proportion of users who complete a purchase after clicking on a product, i.e., the ratio of purchases to clicks. For telecom product recommendations, PCR is a key indicator for measuring whether recommended products truly meet user needs and can effectively promote sales. For example, if a product has a high CTR but a relatively low PCR, it may indicate that although users are interested in the product, the likelihood of them ultimately deciding to buy is low. Conversely, if a product has a high PCR-to-CTR ratio, it means that once a user clicks, the probability of buying is high, making such a product more attractive to users. The PCR-to-CTR ratio is used as an additional adjustment factor in the subsequent recommendation scoring calculation, reflecting the actual feedback effect of user behavior, i.e., the user's true response to the recommended products.

[0075] In the optimization of recommendation models, the ratio of purchase conversion rate to click-through rate is used as an additional adjustment coefficient to adjust the recommendation score and evaluate the quality of the recommendation results. A higher ratio means that users have a stronger actual willingness to purchase the recommended products, reflecting the effectiveness of the product recommendations. By incorporating this ratio into model training, the model parameters can be adjusted to make it pay more attention to products with high conversion rates, thereby optimizing the recommendation strategy. This also ensures that the recommendation algorithm can adapt to changes in user behavior patterns in a timely manner, improving the personalization and timeliness of recommendations, as well as the overall recommendation effect and user conversion rate.

[0076] In this embodiment, by collecting target user attribute information (including age, gender, identity, etc.) and behavioral information (such as purchase and browsing history), combined with detailed attributes and market behavior data of products in the product pool, a pre-trained product recommendation model can generate highly personalized product recommendations. During the training process of the pre-trained product recommendation model, a collaborative filtering algorithm is used to calculate the first similarity between users and the second similarity between products. This not only helps the model identify common preferences within user groups but also understands the correlation between products. By combining similarity with user product interest ratings, user preferences, and product recommendation priorities, the model can generate more optimized recommendation decisions, ensuring that recommended products... Not only does it meet users' personalized needs, but it also possesses high market appeal and potential value. The model design incorporates a dynamic feedback mechanism that learns and adjusts from actual user feedback on recommended products (such as click-through rates and purchase conversion rates). By periodically utilizing this feedback information to iterate model training, it continuously updates model parameters, improving prediction accuracy and recommendation effectiveness. The model not only optimizes product recommendations but also supports the determination of personalized recommendation methods. It can intelligently match the most suitable recommendation channels based on user identity, age, gender, and other information, making the recommendation service more closely aligned with user habits and preferences. This solves the technical problem that related product recommendation schemes struggle to meet users' personalized needs and achieve poor recommendation results.

[0077] Example 2

[0078] According to an embodiment of this application, a product recommendation device for implementing the product recommendation method in Embodiment 1 is also provided, such as... Figure 2 As shown, the product recommendation device includes at least: an acquisition module 21, an analysis module 22, and a recommendation module 23, wherein:

[0079] Module 21 is used to acquire target user attribute information and target user behavior information of the target user, and target product attribute information and target product behavior information of all products in the product pool;

[0080] Analysis module 22 is used to analyze target user attribute information, target user behavior information, target product attribute information, and target product behavior information using a pre-trained product recommendation model to obtain target recommended products corresponding to the target users. The training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined by the following method: using a collaborative filtering algorithm to analyze user attribute information, user behavior information, product attribute information, and product behavior information to obtain the first similarity between multiple users and the second similarity between multiple products, and determining each user's interest rating, user preference, and recommendation priority of each product for each user based on the first and second similarities, and determining the recommended products corresponding to each user based on the interest rating, user preference, and recommendation priority.

[0081] The recommendation module 23 is used to determine the target product recommendation method corresponding to the target user's attribute information, and recommend target recommended products to the target user based on the target product recommendation method.

[0082] The following section explains the functions of each module of the product recommendation device in conjunction with the specific implementation process.

[0083] The acquisition module obtains the target user's attribute information and target user behavior information, as well as the target product's attribute information and target product behavior information for all products in the product pool. By gaining a deeper understanding of the characteristics of the target user and all products in the product pool, it generates more accurate product recommendations.

[0084] As an optional implementation, user attribute information includes at least one of the following: user's age, gender, address, and services already processed; user behavior information includes at least one of the following: user's browsing history, purchase history, search history, and rating history; product attribute information includes at least one of the following: product type, price, brand, and network standard; product behavior information includes at least one of the following: product browsing history, purchase history, search history, and rating history.

[0085] After obtaining the above information, the analysis module uses a pre-trained product recommendation model to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information to obtain the target recommended products corresponding to the target users. The training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined in the following way: the collaborative filtering algorithm is used to analyze the user attribute information, user behavior information, product attribute information, and product behavior information to obtain the first similarity between multiple users and the second similarity between multiple products. Based on the first similarity and the second similarity, the interest rating, user preference, and recommendation priority of each product for each user are determined. Based on the interest rating, user preference, and recommendation priority, the recommended products corresponding to each user are determined.

[0086] As an optional implementation method, the training process of the product recommendation model can be carried out in the following steps:

[0087] S1, Construct an initial recommendation model based on deep learning;

[0088] S2, obtain user attribute information and user behavior information of multiple users, product attribute information and product behavior information of multiple products, and use the user attribute information and user behavior information of each user, and the product attribute information and product behavior information of multiple products as a set of training samples, and determine the sample label corresponding to each set of training samples.

[0089] S3 uses multiple sets of training samples and sample labels to iteratively train the initial recommendation model, and uses the stochastic gradient descent method to optimize the model parameters. When the preset training termination condition is met, the trained model is used as the product recommendation model.

[0090] As an optional implementation, during model training, the following process can be used to determine the sample labels corresponding to each group of training samples:

[0091] S1. Construct a user-product co-occurrence matrix based on user behavior information of multiple users and product behavior information of multiple products. The elements in the user-product co-occurrence matrix are the comprehensive ratings of each user for each product. The comprehensive rating is the score calculated by weighting the number of different behaviors of each user for each product and the rating scores.

[0092] As an optional implementation method, the formula for calculating the overall score is as follows:

[0093] r ui =ω1·b1+ω2·b2+ω3·b3+ω4·b4+ω5·s4

[0094] In the formula, r ui This represents the overall rating of user u for product i. ω1, ω2, ω3, ω4, and ω5 are preset weighting coefficients, and b1, b2, b3, b4, and s4 represent the number of times user u viewed, purchased, searched, rated, and rated product i, respectively.

[0095] S2, based on the user-item co-occurrence matrix, uses a user-based collaborative filtering algorithm to determine the first similarity between every two users and the first interest rating of each user for each item.

[0096] The formula for calculating the first similarity is:

[0097]

[0098] In the formula, sim(u,v) represents the first similarity between user u and user v, A represents the set of items for which both user u and user v have performed actions, and r uk This represents user u's overall rating of product k, r vk This represents the overall rating of user v for product k, ∑ k∈A r uk ·r vk This represents the dot product of user u and user v's ratings on all items k. The larger the dot product, the more similar the ratings of the two users on these items. Let Euclidean norm (i.e., the length of the vector) represent the rating vectors of user u and user v. By dividing by these two norms, the dot product is normalized to a value between -1 and 1. This is the cosine similarity. The closer the cosine similarity is to 1, the more similar the behaviors of the two users are.

[0099] The formula for calculating the first interest rating is:

[0100]

[0101] In the formula, P ui1 Let denot sim(u,n) represent user u's first interest rating for product i. This rating is predicted based on the ratings of other users similar to user u for product i. In other words, by finding other users with similar behavior to user u and inferring user u's potential interest in product i based on their ratings, N(u) represents the set of N users with the highest first similarity to user u, and sim(u,n) represents the first similarity between user u and user n. ni This represents the overall rating of user n for product i.

[0102] S3 uses a product-based collaborative filtering algorithm to determine the second similarity between every two products and the second interest rating of each user for each product;

[0103] The formula for calculating the second similarity is:

[0104]

[0105] In the formula, sim(i,j) represents the second similarity between product i and product j, B represents the set of users who have interacted with both product i and product j, and r ni R represents the overall rating of user n for product i. nj This represents user n's overall rating of product j.

[0106] The formula for calculating the second interest rating is:

[0107]

[0108] In the formula, P ui2 Let sim(i,k) represent the second interest rating of user u for product i, N(i) represent the set of N products with the highest second similarity to product i, sim(i,k) represent the second similarity between product i and product k, and r uk This represents user u's overall rating of product k.

[0109] S4. For each user and each product, determine the user's preference coefficient for the product based on the first user rating and the first user similarity. The first user rating is the comprehensive rating of the user for the product in the user-product co-occurrence matrix, and the first user similarity is the average of the first similarity of the user with all other users who have acted on the product.

[0110] The formula for calculating the preference coefficient is:

[0111]

[0112] In the formula, α ui r represents the preference coefficient of user u for product i. ui Let represent the overall rating of user u for product i, C represent the set of users other than user u who have interacted with product i, sim(u,n) represent the first similarity between user u and user n, and c1 and c2 are preset constants;

[0113] S5. Determine the recommendation priority coefficient of the product to the user based on the first user rating, the first product rating, the first user similarity, and the first product similarity. The first product rating is the average of the ratings of all users to the product in the user-product co-occurrence matrix, and the first product similarity is the average of the second similarities between the product and other products that have generated user behavior.

[0114] The formula for calculating the recommendation priority coefficient is:

[0115]

[0116] In the formula, β ui Let represent the recommendation priority coefficient of product i for user u, C represent the set of other users besides user u who have interacted with product i, sim(u,n) represent the first similarity between user u and user n, D represent the set of other products besides product i that user u has interacted with, sim(i,k) represent the second similarity between product i and product k, and r ui Let r represent the overall rating of user u for product i, E represent the set of users who have interacted with product i, and r represent the overall rating of user u for product i. ni This represents the overall rating of user n for product i, where c3 and c4 are preset constants;

[0117] S6 determines the product's recommendation rating for the user based on the preference coefficient, recommendation priority coefficient, first interest rating, and second interest rating.

[0118] The formula for calculating the recommendation rating is:

[0119] S ui =α ui ·β ui ·θ1·P ui1 ·θ2·P ui2

[0120] In the formula, S ui α represents the recommendation rating of product i for user u. ui β represents the preference coefficient of user u for product i. ui P represents the recommendation priority coefficient of product i for user u. ui1 P represents user u's first interest rating for product i. ui2 θ1 and θ2 represent the second interest rating of user u for product i, and are preset weight coefficients.

[0121] S7. For each user, sort multiple products from highest to lowest based on the user's recommendation ratings for multiple products, determine the top-ranked preset number of products as the user's recommended products, and use the recommended products as the sample labels for the user's corresponding training samples.

[0122] After training the product recommendation model using the above method, the model is used to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information to obtain the target recommended products corresponding to the target user.

[0123] After obtaining the target recommended products corresponding to the target user, the recommendation module determines the target product recommendation method corresponding to the target user's attribute information, and recommends the target recommended products to the target user based on the target product recommendation method.

[0124] The target user attribute information includes the target user's identity, age, and gender. The process of determining the target product recommendation method corresponding to the target user's attribute information can take the following steps: Query the product recommendation method list for the product recommendation method corresponding to the target user's identity; the product recommendation method list stores product recommendation methods corresponding to different user identities; if a product recommendation method exists corresponding to the target user's identity, use that product recommendation method as the target product recommendation method; if no product recommendation method exists corresponding to the target user's identity, determine a template recommendation method from the product recommendation method list that matches the target user's age and gender as the target product recommendation method; in response to the target user's adjustment of the template recommendation method, store the adjusted template recommendation method as the product recommendation method corresponding to the target user's identity in the product recommendation method list.

[0125] As an optional implementation, after recommending target products to target users based on the target product recommendation method, the following steps can be taken to further collect feedback data from target users on the recommended products in order to optimize the performance of the recommendation model:

[0126] Obtain the click-through rate and purchase conversion rate of target users for target recommended products, and determine the ratio of purchase conversion rate to click-through rate; use target user attribute information, target user behavior information, target product attribute information, and target product behavior information as new training samples, and use the ratio as an additional adjustment coefficient when determining the recommendation rating of the target recommended product for the target user during the process of determining the sample labels of the new training samples; periodically use the new training samples and corresponding sample labels to iteratively train the product recommendation model and update the model parameters.

[0127] By incorporating actual user behavior feedback (click-through rate and purchase conversion rate), the system can periodically adjust and optimize the product recommendation model. This mechanism ensures that the recommendation algorithm can adapt to changes in user behavior patterns in a timely manner, improving the personalization and timeliness of recommendations. Simultaneously, by using user feedback as an additional adjustment factor, the model can more accurately predict users' interests in products, thereby improving the conversion rate of recommended products and user satisfaction.

[0128] It should be noted that each module in the product recommendation device in this application embodiment corresponds one-to-one with each implementation step of the product recommendation method in embodiment 1. Since embodiment 1 has been described in detail, some details not shown in this embodiment can be referred to embodiment 1, and will not be elaborated further here.

[0129] Example 3

[0130] According to an embodiment of this application, a computer program product is also provided, which includes a computer program, wherein the computer program, when executed by a processor, implements the product recommendation method in Embodiment 1.

[0131] According to an embodiment of this application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device containing the non-volatile storage medium executes the product recommendation method in Embodiment 1 by running the computer program.

[0132] According to an embodiment of this application, a processor is also provided for running a computer program, wherein the computer program executes the product recommendation method in Embodiment 1 during runtime.

[0133] According to an embodiment of this application, an electronic device is also provided, comprising: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the product recommendation method of Embodiment 1 through the computer program.

[0134] Specifically, the computer program executes the following steps during runtime: acquiring target user attribute information and target user behavior information, and target product attribute information and target product behavior information for all products in the product pool; analyzing the target user attribute information, target user behavior information, target product attribute information, and target product behavior information using a pre-trained product recommendation model to obtain target recommended products corresponding to the target user. The training samples for training the product recommendation model include: user attribute information and user behavior information for multiple users, and product attribute information and product behavior information for multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined in the following way: analyzing user attribute information, user behavior information, product attribute information, and product behavior information using a collaborative filtering algorithm to obtain the first similarity among multiple users and the second similarity among multiple products; determining each user's interest rating, user preference, and recommendation priority for each product based on the first and second similarities; determining the recommended products corresponding to each user based on the interest rating, user preference, and recommendation priority; determining the target product recommendation method corresponding to the target user attribute information, and recommending target recommended products to the target user based on the target product recommendation method.

[0135] As an alternative implementation, the above-mentioned electronic device may exist in the form of a mobile terminal, a computer terminal, or a similar computing device. Figure 3 A hardware block diagram of an electronic device for implementing a product recommendation method is shown. Figure 3 As shown, the electronic device 30 may include one or more processors 302 (shown as 302a, 302b, ..., 302n in the figure) (processor 302 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 304 for storing data, and a transmission device 306 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 3 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, electronic device 30 may also include... Figure 3 The more or fewer components shown, or having the same Figure 3 The different configurations shown.

[0136] It should be noted that the aforementioned one or more processors 302 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element of the electronic device 30. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0137] The memory 304 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the product recommendation method in this embodiment. The processor 302 executes various functional applications and data processing by running the software programs and modules stored in the memory 304, thereby implementing the aforementioned application vulnerability detection method. The memory 304 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 304 may further include memory remotely located relative to the processor 302, and these remote memories can be connected to the electronic device 30 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0138] The transmission device 306 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 30. In one example, the transmission device 306 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 306 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0139] The display may be, for example, a touchscreen liquid crystal display (LCD), which allows the user to interact with the user interface of the electronic device 30.

[0140] The sequence numbers of the above embodiments are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0141] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0142] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0143] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0144] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0145] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0146] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A product recommendation method, characterized in that, include: Obtain target user attribute information and target user behavior information for the target user, and target product attribute information and target product behavior information for all products in the product pool; A pre-trained product recommendation model is used to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information to obtain target recommended products corresponding to the target user. The training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined in the following manner: A collaborative filtering algorithm is used to analyze the user attribute information, user behavior information, product attribute information, and product behavior information to obtain a first similarity among the multiple users and a second similarity among the multiple products. Based on the first similarity and the second similarity, each user's interest rating for each product, user preference, and recommendation priority for each product to each user are determined. Finally, the recommended products corresponding to each user are determined based on the interest rating, user preference, and recommendation priority. Determine the target product recommendation method corresponding to the target user attribute information, and recommend the target recommended product to the target user based on the target product recommendation method; The process of determining the interest rating, user preferences, and recommendation priority includes: A user-product co-occurrence matrix is ​​constructed based on the user behavior information of the multiple users and the product behavior information of the multiple products. The elements in the user-product co-occurrence matrix are the comprehensive ratings of each user for each product. Based on the user-item co-occurrence matrix, a user-based collaborative filtering algorithm is used to determine the first similarity between each pair of users and the first interest rating of each user for each item. An item-based collaborative filtering algorithm is used to determine the second similarity between each pair of items and the second interest rating of each user for each item. For each user and each product, the user's preference for the product is determined based on a first user rating and a first user similarity, wherein the first user rating is the comprehensive rating of the user for the product in the user-product co-occurrence matrix, and the first user similarity is the average of the first similarities of the user and all other users who have acted on the product; The recommendation priority of the product to the user is determined based on the first user rating, the first product rating, the first user similarity, and the first product similarity. The first product rating is the average of the ratings of all users for the product in the user-product co-occurrence matrix, and the first product similarity is the average of the second similarities between the product and other products that have generated the user's behavior.

2. The method according to claim 1, characterized in that, The training process of the product recommendation model includes: Construct an initial recommendation model based on deep learning; The system acquires user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. It then uses the user attribute information and user behavior information of each user, and the product attribute information and product behavior information of the multiple products as a set of training samples, and determines the sample label corresponding to each set of training samples. The initial recommendation model is iteratively trained using multiple sets of training samples and sample labels, and the model parameters are optimized using the stochastic gradient descent method. When the preset training termination condition is met, the trained model is used as the product recommendation model.

3. The method according to claim 2, characterized in that, The comprehensive score is a weighted score calculated by summing the number of different behaviors and evaluation scores of each user for each product. The sample labels corresponding to each group of training samples are determined, including: For each user and each product, the recommendation rating of the product for the user is determined based on the user's user preference for the product, the recommendation priority of the product for the user, the user's first interest rating and second interest rating for the product; For each user, the multiple products are sorted from largest to smallest based on the user's recommendation rating for the multiple products. A preset number of products at the top of the sorted list are determined as the recommended products for the user, and these recommended products are used as the sample labels for the training samples corresponding to the user.

4. The method according to claim 3, characterized in that, The formula for calculating the comprehensive score is as follows: r ui =ω1·b1+ω2·b2+ω3·b3+ω4·b4+ω5·s4 In the formula, r ui This represents the overall rating of user u for product i. ω1, ω2, ω3, ω4, and ω5 are preset weighting coefficients, and b1, b2, b3, b4, and s4 represent the number of times user u viewed, purchased, searched, rated, and rated product i, respectively. The formula for calculating the first similarity is: In the formula, sim(u, v) represents the first similarity between user u and user v, A represents the set of items for which both user u and user v have interacted, and r uk This represents user u's overall rating of product k, r vk This represents user v's overall rating of product k; The formula for calculating the first interest score is: In the formula, P ui1 Let sim(u,n) represent the first interest rating of user u for product i, N(u) represent the set of N users with the highest first similarity to user u, and r represent the first similarity between user u and user n. ni This represents user n's overall rating of product i; The formula for calculating the second similarity is: In the formula, sim(i,j) represents the second similarity between product i and product j, B represents the set of users who have interacted with both product i and product j, and r ni R represents the overall rating of user n for product i. nj This represents user n's overall rating of product j; The formula for calculating the second interest rating is: In the formula, P ui2 Let sim(i, k) represent the second interest rating of user u for product i, N(i) represent the set of N products with the highest second similarity to product i, sim(i, k) represent the second similarity between product i and product k, and r uk This represents user u's overall rating of product k; The formula for calculating user preferences is: In the formula, α ui r represents user u's preference for product i. ui Let represent the overall rating of user u for product i, C represent the set of users other than user u who have interacted with product i, sim(u,n) represent the first similarity between user u and user n, and c1 and c2 are preset constants; The formula for calculating the recommendation priority is: In the formula, β ui Let represent the recommendation priority of product i for user u, C represent the set of other users besides user u who have interacted with product i, sim(u, n) represent the first similarity between user u and user n, D represent the set of other products besides product i that user u has interacted with, and sim(i, k) represent the second similarity between product i and product k. ui Let r represent the overall rating of user u for product i, E represent the set of users who have interacted with product i, and r represent the overall rating of user u for product i. ni This represents the overall rating of user n for product i, where c3 and c4 are preset constants; The formula for calculating the recommendation score is as follows: S ui =a ui ·b ui ·θ1·P ui1 ·θ2·P ui2 In the formula, S ui α represents the recommendation rating of product i for user u. ui β represents user u's preference for product i. ui P represents the recommendation priority of product i to user u. ui1 P represents user u's first interest rating for product i. ui2 θ1 and θ2 represent the second interest rating of user u for product i, and are preset weight coefficients.

5. The method according to claim 1, characterized in that, The target user attribute information includes: the target user's identity identifier, age, and gender. Determining the target product recommendation method corresponding to the target user attribute information includes: The product recommendation method is retrieved from the product recommendation method list to match the identity of the target user. The product recommendation method list stores product recommendation methods corresponding to the identity of different users. If there is a product recommendation method that corresponds to the identity identifier of the target user, the product recommendation method shall be used as the target product recommendation method; If no product recommendation method corresponds to the target user's identity, a template recommendation method matching the target user's age and gender is determined from the product recommendation method list and used as the target product recommendation method; in response to the target user's adjustment of the template recommendation method, the adjusted template recommendation method is stored as the product recommendation method corresponding to the target user's identity in the product recommendation method list.

6. The method according to claim 3, characterized in that, After recommending the target recommended product to the target user based on the target product recommendation method, the method further includes: Obtain the click-through rate and purchase conversion rate of the target user for the target recommended product, and determine the ratio of the purchase conversion rate to the click-through rate; The target user attribute information, the target user behavior information, the target product attribute information, and the target product behavior information are used as a new training sample. In the process of determining the sample label of the new training sample, when determining the recommendation rating of the target recommended product for the target user, the ratio is used as an additional adjustment coefficient to determine the recommendation rating of the target recommended product for the target user. The product recommendation model is periodically trained using new training samples and corresponding sample labels to update the model parameters.

7. The method according to claim 1, characterized in that, The user attribute information includes at least one of the following: the user's age, gender, address, and transactions already processed; The user behavior information includes at least one of the following: the user's browsing history, purchase history, search history, and rating history; The product attribute information includes at least one of the following: product type, price, brand, and network standard; The product behavior information includes at least one of the following: product browsing history, purchase history, search history, and rating history.

8. A product recommendation device, characterized in that, include: The acquisition module is used to acquire target user attribute information and target user behavior information of the target user, and target product attribute information and target product behavior information of all products in the product pool. The analysis module is used to analyze the target user attribute information, target user behavior information, target product attribute information, and target product behavior information using a pre-trained product recommendation model to obtain target recommended products corresponding to the target user. The training samples for training the product recommendation model include: user attribute information and user behavior information of multiple users, and product attribute information and product behavior information of multiple products. The sample labels for training the product recommendation model are the recommended products corresponding to each user, determined by the following method: using a collaborative filtering algorithm to analyze the user attribute information, user behavior information, product attribute information, and product behavior information to obtain a first similarity among the multiple users and a second similarity among the multiple products; determining each user's interest rating, user preference, and recommendation priority for each product based on the first and second similarities; and determining the recommended products corresponding to each user based on the interest rating, user preference, and recommendation priority. The process of determining the interest rating, user preference, and recommendation priority includes: based on the user behavior information of the multiple users and the product... A user-item co-occurrence matrix is ​​constructed using behavioral information, where each element in the matrix is ​​a user's overall rating for each item. Based on this matrix, a user-based collaborative filtering algorithm is used to determine the first similarity between any two users and the first interest rating of each user for each item. Similarly, an item-based collaborative filtering algorithm is used to determine the second similarity between any two items and the second interest rating of each user for each item. For each user and each item, the user's preference for the item is determined based on the first user rating and the first user similarity, where the first user rating is the user's overall rating for the item in the user-item co-occurrence matrix, and the first user similarity is the average of the first similarities between the user and all other users who have interacted with the item. Finally, the recommendation priority of the item for the user is determined based on the first user rating, the first item rating, the first user similarity, and the first item similarity, where the first item rating is the average of all user ratings for the item in the user-item co-occurrence matrix, and the first item similarity is the average of the second similarities between the item and other items interacting with the user. The recommendation module is used to determine the target product recommendation method corresponding to the target user attribute information, and recommend the target recommended product to the target user based on the target product recommendation method.

9. A computer program product, characterized in that, include: A computer program, wherein when executed by a processor, the computer program implements the product recommendation method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the product recommendation method according to any one of claims 1 to 7 through the computer program.

Images